









































Pa
ge

 
1



Pa
ge

 
10

3

American Journal of  Medical 
Science and Innovation (AJMSI) 

Digital Revolution in Medical Pathology: Integrating Ai, Genomics, and Molecular Imaging
Adeyemi Sarah Halleluyah1*, Sodiq Murphy Balogun2, Abraham Chibuikem Ikeji3, Bukola E. Shasere3, Omeshamisu Anigala4

Volume 4 Issue 2, Year 2025
ISSN: 2836-8509 (Online)

DOI: https://doi.org/10.54536/ajmsi.v4i2.5281
https://journals.e-palli.com/home/index.php/ajmsi

Article Information ABSTRACT

Received: May 22, 2025
Accepted: June 26, 2025
Published: November 10, 2025

The current clinical pathologic diagnostic process using histological slide evaluation 
demonstrates inconsistent accuracy in medical diagnosis. Progress in digital pathology and 
other modern technologies now enables the utilization of  AI for medical image diagnostics, 
along with genomic technology for profile assessment and molecular image functionality. 
This systematic review examines the impact of  artificial intelligence technology combined 
with genomic analysis and molecular imaging systems on present-day pathological medicine 
advancement. The review includes research published between 2014 and 2025, obtained from 
the top five databases, to demonstrate how each technology improves diagnosis separately 
and collaborates for precise medicine advancement. The analysis evaluated ten studies that 
matched all the established criteria for inclusion. Medical diagnostics benefit from combined 
system platforms, which also strengthen patient classification systems and treatment 
selection. However, these platforms require improvements in data standards and workflow 
connections, as well as computational resources and moral framework requirements. The 
study defines the necessary criteria for government-connected data platforms and interpretive 
artificial intelligence models, and then creates regulatory mechanisms in collaboration with 
interdisciplinary partnerships to develop safe and equitable healthcare applications. A single 
organized system triggers an essential transformation that shifts pathology from traditional 
morphological practices toward complex multivariate modern data methods. The research 
delivers strategic recommendations to enhance future practice and policy development, 
which will enable these technologies to be widely used in clinical settings.

Keywords
Artificial Intelligence, Digital 
Pathology, Genomics, Molecular 
Imaging,  Precision Medicine

1 Bioinformatics, Morgan State University, Maryland, USA
2 Department of  Bioinformatics and Genomics, University of  North Carolina at Charlotte, USA
3 Mayo Clinic, USA
4 Department of  Electrical Engineering and Computer Science South Dakota State University, USA
* Corresponding author’s e-mail: sarahadeyemi362@gmail.com

INTRODUCTION
The medical field of  pathology maintains crucial 
importance for disease examination, which supports both 
diagnosis and treatment of  patients (Ahuja & Zaheer, 
2025). The pathological approach of  tissue section 
examination under a microscope faces two drawbacks: 
it depends on subjective visual interpretation, and it 
produces variable results between different observers 
according to Madabhushi & Lee (2016). The diagnostic 
field of  medicine underwent important changes in recent 
times because innovative technologies seek to enhance 
diagnostic accuracy, combined with quicker operations.  
Digital pathology brought about significant changes in 
healthcare through vigorous digitization of  medical slides 
while creating opportunities for remote consultations 
and image analysis, and storing large amounts of  
valuable data (Ali & Saqib, 2023). Digital transformation 
has optimized operational processes and established 
conditions for advanced computational systems to enter 
modern pathological procedures (Shafi & Parwani, 
2023). Computational algorithms linked with digital 
imaging created new methods to analyze information 
quantitatively which leads to more dependable 
measurement systems. The incorporation of  Artificial 
Intelligence (AI) with genomics and molecular imaging 
into pathology practice leads scientific advancements 
in delivering precision medicine (Munari et al., 2024). 

Artificial Intelligence delivers outstanding performance 
through deep learning algorithms that outstrips human 
competency during specific diagnostic examinations 
(Sussman et al., 2022). Digital histopathological image 
analysis through AI models enables medical experts to 
detect different cancers (Abasher  et al., 2023), while 
measuring treatment effects and forecasting patient 
clinical results according to Salo et al. (2024). Massive data 
training enables such systems to detect faint patterns seen 
only by sophisticated machines which results in improved 
diagnostic accuracy. The molecular brightness of  diseases 
has advanced significantly because of  genomics (Ikwuka 
et al., 2013). Modern sequencing methods provide full 
genetic alteration exploration capabilities which lead 
to biomarker discoveries for diagnostic testing and 
prognosis prediction and therapeutic aims (Munari et 
al., 2024). The combination of  genomic data analysis 
with tissue examination results enables detailed disease 
classification, especially in cancer cases, which leads 
healthcare providers to design individualised treatments 
(Asif  et al., 2023). Treatment effect, along with patient 
results improves with targeted therapy decisions based 
on specific genomic profile mutations. Through positron 
emission tomography (PET) and single-photon emission 
computed tomography (SPECT), which belong to 
molecular imaging technologies, researchers can observe 
functional biological processes within living patients. 



Pa
ge

 
10

4

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

Medical experts use these diagnostic methods to monitor 
receptor expression levels along with metabolic activity by 
gaining additional secondary information from traditional 
anatomical imaging data (Fahmy, 2024). The combination 
of  molecular imaging with AI algorithm enhancements 
facilitates better pathological detection alongside better 
pathological variation characterization through complex 
detection methods and exact identification (Rowe et al., 
2021). Better disease assessment capabilities within this 
combined system led to immediate, proper medical 
interventions.

Rationale for the Study
Pathological applications experience very limited AI 
advancements because genomic and molecular-imaging 
technologies fail to achieve sufficient collaborative 
development. Modern technologies demonstrate the 
capability to develop morphology-based pathology 
into an advanced data-based field that extends beyond 
traditional morphology-based practices. Better disease 
knowledge combined with superior patient outcomes 
becomes achievable when pathologists analyze disease 
observations together with genetic and molecular data 
according to Munari et al. (2024). The integration process 
reveals different challenges because it needs standardized 
data formats alongside system interoperability and training 
for specialized staff  members. The implementation 
of  these tools in healthcare facilities requires proper 
solutions to both data privacy concerns and algorithm 
transparency needs to sustain ethical clinical practice (Asif  
et al., 2023). Healthcare providers and scientists alongside 
policy experts need to unite their efforts for developing 
standardized rules which defend patient safety and enable 
these tools to function in typical medical care.

Research Aim and Objectives
This study aims to explore the transformative impact of  
integrating AI, genomics, and molecular imaging into 
medical pathology. The specific objectives are:

1. To review the current applications and advancements 
of  AI in pathological diagnostics.

2. To examine the role of  genomic data in enhancing 
pathological assessments.

3. To evaluate the contributions of  molecular imaging 
techniques in pathology.

4. To identify the challenges and limitations associated 
with the integration of  these technologies.

5. To propose recommendations for effective 
implementation and future research directions.

MATERIALS AND METHODS
Study Design 
The review implemented PRISMA guidelines throughout 
its systematic methodology (Page et al., 2021). The 
research design brings together established studies 
about AI and genomic analysis with molecular imaging 
in pathology to achieve both less subjective decision-
making and enhanced research reproducibility. The 

study used two independent reviewers who conducted 
dual screenings for data extraction concurrently until 
a third expert resolved any discrepancies. The review 
protocol established both eligibility requirements and 
data collection items and quality assessment instruments 
(CASP and Newcastle–Ottawa Scale) before starting the 
search process to avoid post-hoc decision-making. The 
PRISMA framework enhances research methodology, 
but is still unable to eliminate both publication bias and 
differences among research approaches. The system of  
registered protocols depends on certain assumptions 
about database availability, but does not identify research 
that exists outside database systems. The use of  systematic 
registration during complete evidence synthesis helps 
researchers execute established guidelines correctly.

Inclusion Criteria 
This study analyzed peer-reviewed original articles 
which met the following four conditions: (1) used AI 
algorithms with genomic analysis or molecular imaging 
applications in human pathology settings, (2) included 
clinical or histopathological specimen data such as 
biopsies and resection specimens, (3) provided diagnostic 
performance data together with workflow influences 
and patient-centered outcomes and (4) were published 
in English during January 2014 to March 2025. The 
analysis included clinical trials and cohort studies as 
well as case-control investigations and cross-sectional 
research and technology-validation studies for collecting 
evidence at multiple levels. The scope supports practical 
applications in medical practices through its multiple 
research method acceptance. The review only accepting 
English-language publications could potentially hide 
groundbreaking research from other languages which 
results in linguistic bias affecting the study results. The 
ten‐year time boundary safeguards contemporary digital 
innovation studies without undermining earlier research 
conducted between 2014 and the present day. The 
research lacked geographic limitations letting participants 
from worldwide locations submit data yet this method 
brought increased variability among different healthcare 
facilities and their available resources.

Exclusion Criteria
The review excluded (1) non–peer‐reviewed literature 
(editorials, commentaries, conference abstracts, theses), 
to focus on fully vetted research; (2) studies without direct 
pathology relevance (e.g., radiology‐only AI applications 
or bioinformatics pipelines lacking histopathological 
correlation); (3) purely in vitro or animal‐model 
investigations without human data; (4) articles lacking 
sufficient methodological detail or performance metrics; 
and (5) duplicates and extensions of  the same primary 
dataset. The chosen data restriction criteria improve 
data review precision and quality but removes essential 
early-stage research that often presents at conferences. 
Language bias results from reports written in any 
language except English while publication bias becomes 



Pa
ge

 
10

5

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

more prominent when gray literature is excluded. The 
team documented all excluded studies for potential use in 
future research updates that may include different types 
of  evidence during the field’s development.

Search Strategy 
A systematic method was used for comprehensive 
literature search that combined Boolean operators along 
with Medical Subject Headings (MeSH) and free-text 
terms. The search terms were chosen specifically to 
represent digital pathology and artificial intelligence (AI) 
together with genomics and molecular imaging domains. 
The search included the following grouping of  terms: 
“Digital pathology” OR “Whole slide imaging” along 
with “Artificial Intelligence” OR “AI in pathology” 
OR “Machine learning” OR “Deep learning” as well as 
“Histopathology” OR “Tissue analysis” and “Genomics” 
OR “Next-generation sequencing” OR “Genetic 
profiling” plus “Molecular imaging” OR “PET” OR 
“SPECT” together with “Precision medicine” AND 
“Pathology”. The researchers used Boolean operators 
(AND, OR) to properly connect their selected concepts. 
For example: (“Digital pathology” AND “Artificial 
Intelligence”) OR (“Genomics” AND “Histopathology”) 
AND (“Molecular imaging” OR “Deep learning”). 
The literature search included PubMed/MEDLINE 
and Scopus and Web of  Science and IEEE Xplore to 
complement Google Scholar. English-language peer-
reviewed articles published since January 1, 2014 until 
March 31, 2025 made up the scope of  this research. The 
authors performed their screening by PRISMA guidelines. 
The EndNote X9 program imported the search results for 
duplicate detection. Two separate researchers evaluated 
titles and abstracts of  potential studies to determine 
their eligibility. Two researchers performed detailed 
reviews of  articles that showed potential relevance. The 
research team excluded studies which failed to match the 
selection criteria while providing documented reasons. 
Any disagreement between reviewers was settled through 
mutual discussion or involvement of  a third-party 
arbitrator. A systematic process followed by transparent 
methods enabled researchers to capture high-quality 
relevant studies that answered the study objectives.

Data Extraction and Management
A standardized extraction form was tested for clarity 
and consistency using ten randomly chosen medical 
studies before implementation on the remaining studies. 
The data collection process acquired information about 
study authors, publication dates, countries of  origin as 
well as research design types and sample sizes, histology 
and cytology methods, AI system frameworks, genomic 
sequencing techniques, molecular imaging tracers, 
performance metrics, and documented clinical and 
workflow impact metrics. The extraction process took 
place independently between two reviewers who used 
Microsoft Excel software with automatic version tracking 
capabilities. The reviewers checked all measurements 

showing more than 10% deviation and met to reach 
consensus; disagreements escalated to a third expert 
validation.

Prisma
Transparent and rigorous selection followed the PRISMA 
2020 guidelines during the study assessment process. The 
search process identified 1,374 records, which included 
database search results combined with manual reference 
tracking. The database search yielded 410 articles from 
PubMed and 340 from Scopus, together with 290 from 
Web of  Science and 157 from IEEE Xplore and 177 
from Google Scholar. Through the process of  duplicate 
removal, 1,062 unique records persisted. Two independent 
reviewers reviewed titles and abstracts, which resulted 
in discarding 931 articles due to their irrelevance to the 
research topic and their non-human data or lack of  digital 
pathology technology focus. A total of  131 full articles 
underwent a methodological assessment as well as a 
relevance review for their connection to AI, genomics, and 
molecular imaging applications in pathology. The analysis 
process excluded 121 articles because the research did not 
integrate the three technologies properly or lacked clinical 
applications or presented methodological issues. A total 
of  ten high-quality studies fulfilled all requirements and 
became part of  the systematic review evaluation. The 
authors conducted a critical assessment of  the included 
studies, which followed thematic synthesis.

Methodology
This systematic review performed a synthesis of  findings 
extracted from chosen research studies during data 
analysis. The research process included extensive database 
searches, which led to selecting relevant studies according 
to established eligibility standards about AI and genomic 
applications as well as molecular imaging in medical 
pathology. A structured data extraction process collected 
essential information, including research designs as 
well as technological methods, measured outcomes and 
application settings. The researchers separated studies 
according to their primary subjects, which examined AI 
diagnostic tool advancements and genomics applications 
in pathology and molecular imaging technique innovations. 
The research team analyzed each theme to detect typical 
patterns alongside current obstacles and new findings 
within this academic subject. A narrative synthesis 
method allowed researchers to organize and make sense 
of  the study results while establishing relationships 
between different research works to present an extensive 
summary of  medical pathology’s up-to-date technologies. 
Researchers designed this study to demonstrate the 
expected improvements which these technologies 
would bring to diagnostic precision and patient recovery 
outcomes, and future pathology operations. A critical 
evaluation of  the selected reports examined both study 
methodology and participant numbers alongside research 
design to guarantee that this review presents validated 
and dependable findings.



Pa
ge

 
10

6

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

RESULTS AND DISUSSION
Technological Innovations and Diagnostic Performance
The combination of  artificial intelligence with genomics 
and molecular imaging technologies enables precision 
diagnostics to transform pathology services by improving 
both diagnostic precision and individualized treatment 
of  patients. Both significant potential advantages and 
critical assessment of  strengths and weaknesses, together 
with combined implementation challenges, must be 
thoroughly evaluated. Such analysis demonstrates how 
complex these systems become when they are put into 
medical practice. Deep learning–based image analysis has 
proven effective in histopathology through Panayides et 
al. (2020) since their method reached area-under-curve 
(AUC) values upward of  0.90 for tumour recognition. 
The obtained results demonstrate AI systems match 
pathologists’ accuracy rates for sensitivity and specificity 
levels. The study demonstrates a major drawback because 
these models fail to maintain consistent performance 
when operating with diverse institutions or changes in 
staining protocols or slide quality (Cheng et al., 2021). 
Large-scale generalization proves to be a major obstacle 
since controlled environment models struggle to perform 
correctly in actual clinical settings. Many AI algorithms 
maintain an untraceable decision-making process, which 
impedes clinicians from adopting them in practice settings 
(Prevedello et al., 2019). The reliability of  AI in pathology 
diagnostic settings comes into question because of  such 
concerns, especially during critical medical determination 
moments. The diagnostic field has experienced significant 
transformation through genomic profiling because it 
identifies mutations and expression patterns to provide 
molecular insights. Tumour stratification in oncology, 
together with predictive treatment responses, increased 
to over 95% sensitivity after integrating next-generation 
sequencing (NGS) data with histopathological images 
according to Seyhan and Carini (2019). The reliability 
of  genomic data faces various challenges according 
to Ahmad et al. (2021). The process of  sequencing 
artifacts together with stringent tumour purity 
requirements can trigger incorrect test results that may 
confuse medical professionals during interpretation. 
Efficient bioinformatics pipelines provide solutions to 
analyse the large quantities of  sequencing data which 
medical practitioners need for clinical applications. The 
bioinformatics pipelines containing AI systems develop 
recursive dependencies, which generate anxiety regarding 
analysis transparency and accumulated mistakes (Seyhan 
& Carini, 2019). Bioinformatics must implement strong, 
transparent procedures to enable genomic data usage in 
clinical decisions (Kermany et al., 2018).
PET and SPECT techniques in molecular imaging 
enable real-time metabolic imaging of  tissue structures 
beneath one centimetre, which traditional histological 
and genomic methods cannot visualise. The research by 
Tian et al. (2021) shows that these imaging modalities 
reach sensitivity rates above 90% for detecting small 
lesions to enhance early detection. PET and SPECT 

imaging methods face difficulties with specificity because 
tracer uptake happens in both malignant and benign 
tissues thereby causing false positive results. Traditional 
microscopic imaging provides higher spatial resolution 
than molecular imaging, which restricts the obtainable 
cellular details from this method. According to Panayides 
et al. (2020) and Simon et al. (2024), accurate interpretation 
and modality synergy between combined analytical 
techniques need precise standardisation of  acquisition 
methods and image processing workflows.
These technological systems, united together, create 
potential major combined advantages. The convergence 
of  image features and mutation profiles, and functional 
imaging within multimodal artificial intelligence systems 
leads to a 15% increase in prognostic accuracy according 
to Simon et al. (2024). The integration process comes 
with significant challenges that need to be addressed. 
The article by Gaffney and Mirza (2025) describes how 
the integration of  these technologies becomes difficult 
due to data format inconsistencies between imaging and 
genomic platforms in addition to complex requirements 
for large dataset management infrastructure and 
divergent governance policies. The matter of  algorithmic 
bias advances as a crucial problem in the field. The 
training of  numerous AI models depends on minority-
underrepresented datasets that result in performance 
differences during clinical use among diverse patient 
populations (Kim et al., 2022).

Workflow Integration and Implementation Barriers 
Focus
The accuracy and efficiency of  routine pathology 
diagnostics are set to improve significantly with 
the integration of  workflow systems that bring 
together AI, genomics, and molecular imaging. These 
technologies encounter multiple complex problems 
while being integrated into current laboratory settings 
which require vital assessment before benefiting from 
their implementation. Shafi and Parwani (2023) state 
that automated whole-slide imaging (WSI) scanners 
introduced digital pathology but their implementation 
requires perfect LIS and reporting platform connectivity. 
Many current LIS systems operate using outdated 
protocols while missing essential APIs which would allow 
them to process large images of  gigapixels and to connect 
genomic reports to histological analysis results. Pathology 
departments struggle with IT infrastructure built for 
transactional reporting which requires extensive hardware 
upgrades of  high-throughput storage area networks and 
comprehensive network modification to handle image and 
sequence data according to Cheng et al. (2021). Panayides 
et al. (2020) emphasize that medical practices need proper 
data management structures based on standardized 
metadata schemas and data lakes to work efficiently. 
Healthcare data remains isolated throughout different 
systems because patients lack common data ontologies 
and tag classification structures (such as DICOM for 
images or HL7 FHIR for clinical information) which 



Pa
ge

 
10

7

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

prevents multi-modality data search and integrated AI 
system development. Federated systems operate without 
centralized raw data storage because algorithms execute 
directly from source nodes to maintain privacy guidelines 
and achieve large-scale algorithms. The entire integrated 
workflow relies on data pipelines which pose the risk 
of  becoming performance bottlenecks. The increase of  
“omics” datasets exceeds statistical capabilities which 
requires sophisticated computational frameworks based 
on Kubernetes clusters for handling large sequencing 
read volumes according to Seyhan and Carini (2019). 
The model performance becomes inconsistent when 
differences in scanner calibration and staining protocols 
and annotation granularity emerge in image acquisition and 
annotation pipelines according to Prevedello et al. (2019).
The use of  centralized image-analysis competitions 
together with shared benchmarking datasets helps with 
both harmonization and ongoing validation process. 
The requirements for implementing this technology 
span technological resources and more. The paper 
explains how research clusters that incorporate GPUs 
or specialized AI accelerators for model training and 
inference also need to include object-storage solutions 
which handle petabyte-scale archives (Kim et al. 2022). The 
acquisition of  capital to fund such infrastructure tends 
to face obstacles against core laboratory budgets which 
leads to resistance from senior management regarding 
investment returns according to Ahmad et al. (2021). The 
necessity for complete cost-benefit assessments emerges 
because reduced turnaround times together with lower 
error rates and possible future savings need to show how 
they balance initial investments. According to Gaffney 
and Mirza (2025) leadership frameworks need to adapt 
simultaneously to create specific guidelines regarding data 
protection together with security protocols and algorithm 
responsibility standards. The implementation of  GDPR 
and HIPAA regulations necessitates role-based access 
controls along with audit trails along with data encryption 
when data rests or when it moves through networks. 
The absence of  proper oversight exposes departmental 
operations to substantial regulatory penalties together with 
a deterioration of  patient confidence. Human factors are 
equally critical. To achieve effective change management 
institutions must provide both tool proficiency training 
to pathologists and technical staff  and education about 
system limitations. The adoption of  AI outputs suffers 
from reduced pathologist acceptance because these 
systems fail to demonstrate their decision processes or 
explain their reasoning according to Cheng et al. (2021). 
The collaboration between IT, bioinformatics and clinical 
teams shows limited success because both departments 
work independently from each other according to Kim 
et al. (2022). Shafi and Parwani (2023) explain that “AI 
champions” who work in pathology laboratories connect 
these two different domains while addressing user 
concerns immediately and facilitate learning between 
colleagues, and drive continuous AI development. 

Successfully adopting AI solutions means providing 
ongoing support, like dedicated help desk staff, regular 
training sessions, and performance dashboards, to keep 
things running smoothly and ensure people continue to 
use the technology effectively.
Panayides et al. (2020) emphasize the necessity for 
integrative analytics platforms to have feedback systems 
that help laboratories enhance their algorithms and 
workflows using actual practice performance indicators. 
Laboratories can achieve reliable routine clinical care 
through proactive protocol adjustments by continuously 
monitoring indicators, which include error rates, model 
drift and user satisfaction (Oala, Flach & Ghalwash, 2022). 

Ethical, Regulatory, and Data Governance 
Considerations  
Combining AI with genomics and molecular imaging in 
pathology opens up incredible opportunities—but it also 
brings significant ethical challenges and regulatory hurdles 
that must be carefully addressed. The research by Gaffney 
and Mirza (2025) shows that patient privacy safeguards, 
together with accountability measures, need equal 
importance to technical performance when diagnostic 
algorithms enter clinical workflows deeply. The absence 
of  strong governance systems will allow sensitive data 
to become compromised or lead to unauthorized uses 
of  confidential information, which would compromise 
technological precision. According to Seyhan and Carini 
(2019), large-scale omics data consolidation poses 
significant threats through combining genomic sequences 
with high-resolution images for AI training models 
because this process centralizes vulnerable patient 
information. Ahmad et al. (2021) emphasise that advanced 
diagnostic democratisation should never violate patient 
consent or autonomy. Panayides et al. (2020) show how 
federated learning addresses this issue through model 
distribution since weight updates get encrypted before 
being shared, according to their research. These security 
systems need perfect encryption with strict key controls 
and accurate access protocols to prevent attackers from 
reassembling private information. The paper written by 
Kim et al. (2022) emphasizes that protecting infrastructure 
stands on par with importance. AI accelerators as well as 
GPUs need protected data centers to operate from, and 
these centres require ongoing monitoring procedures, 
alongside network segmentation, along automated audit 
log generation.
Cheng et al. (2021) emphasize that laboratories must 
comply with frameworks like ISO/IEC 27001 and 
HIPAA because nonexistence of  formal policies for 
encryption and breach notification, and data retention 
leads to regulatory penalties, together with public trust 
deterioration. The majority of  pathology departments 
encounter difficulties converting their legacy IT systems 
which operated for transactional reporting, into systems 
which handle multi-terabyte imaging and sequencing 
archives.



Pa
ge

 
10

8

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

Transparency represents another essential foundation, 
according to the research by Prevedello et al. (2019). The 
secrecy of  “black box” AI models prevents detection of  
scanning artifacts that both break user confidence and 
produce unpredictable mistakes, even when these models 
provide excellent accuracy. Shafi and Parwani (2023) 
explain that pathologists avoid implementing systems that 
lack explainable decision-making capabilities. The study 
from Kim et al. (2022) shows that explainable AI methods 
through attention maps and feature attributions help to 
show prediction formation, yet require accurate micro-
level accuracy and clinical interpretation capabilities. 
The critical dashboard systems described by Panayides 
et al. (2020) help to find both model shifting patterns 
alongside novel biases which negatively impact minority 
populations.
The research conducted by Cheng et al. (2021) reveals 
that regulatory frameworks should advance their rules 
as new technological developments emerge. The FDA 
510(k) clearance system and CE marking approach 
provide regulatory approval for software versions that 
remain unchanged but fail to cover the governance of  
AI systems that learn continuously. Shafi and Parwani 
(2023) highlight recent manufacturer approvals that 
include whole-slide imaging scanners plus a prostate-
cancer AI algorithm, yet create unaddressed questions 
regarding update validation and how to verify retraining, 
along with adaptive learning procedures. Prevedello et 
al. (2019) suggest that mandates for uniform validation 
protocols should include multicenter prospective 
trials and standardised staining along with annotation 
benchmarking to provide regulators with equivalent 
datasets and metrics. The research by Gaffney and Mirza 
(2025) outlines how liability frameworks should specify 
the accountable parties when AI mishaps produce 
injuries. Pathology laboratories need to create oversight 
committees with pharmacovigilance board-like functions 
to verify AI-informed clinical assessments and modify 
consent arrangements for patients, together with adverse 
event tracking. The authors of  Ahmad et al. (2021) 
support the development of  “explainability audits” along 
with continuous post-market surveillance to detect rare 
critical failures. Footprint evaluation depends on clear 
organizational accountability along with rapid information 
feedback between organizations and technicians who 
maintain patient welfare through transparent reporting.

Implication
The research investigation identified multiple essential 
pathologic requirements that require combined technical 
and organizational answers. The highest level of  data 
governance operation must take place. According to 
Seyhan and Carini (2019), the combination of  “omics” and 
imaging data before training AI systems intensifies both 
privacy threats against patient data and potential abusive 
practices related to highly sensitive medical information. 
The federated learning method enables laboratories 
to maintain server-based private data protection and 

encrypted model updates exchange to access multiple 
institutional datasets according to Zhu et al. (2021) and 
Xu et al. (2020). The deployment requires authorised 
policies to establish end-to-end encryption together with 
strict key management protocols and role-based access 
control systems to fulfil the criteria set by ISO/IEC 27001 
and HIPAA standards, according to Cheng et al. (2021) 
and Panayides et al. (2020). AI application infrastructure 
requires design specifications for its functionality. Modern 
healthcare facilities need GPU-powered computing 
clusters and AI accelerator systems that support 
encrypted storage with scale capabilities, according to the 
research by Kim et al. (2022). Modernization efforts must 
be deployed to legacy laboratory information systems, 
which need to use standardized APIs to receive digital 
slides and genomic reports according to Shafi & Parwani 
(2023) and Panayides et al. (2020). Organizations will need 
to show how the combination of  faster test processing 
with fewer mistakes makes the upfront financing expense 
worthwhile (Ahmad et al., 2021). All automated systems 
must implement transparency as an organizational 
foundation. The use of  “Black box” predictive models 
damages trust while it simultaneously conceals any 
random connections that might exist within their results, 
according to Prevedello et al. (2019).
XAI tools with explanatory capabilities, such as HIPPO 
(Arvaniti et al., 2024) as well as attention-based heatmaps 
(Chartrand et al., 2022) provide evaluation functionalities 
for pathologists to analyze models’ decision-making 
processes. Training programs must include specific units 
about understanding AI outputs and their boundary 
limitations (Rai, 2020; Holzinger et al., 2022).
The approval process should evolve from receiving static 
approvals to active oversight. The current FDA 510(k) 
along with CE-mark approval processes, fail to meet 
the requirements of  AI systems that learn in real time. 
Pathology departments need to implement procedures 
for software version control alongside retraining 
validation protocols and post-market safety checks, which 
resemble pharmacovigilance practices (Cheng et al., 2021; 
Prevedello et al., 2019).
Multiple healthcare organizations must establish 
ongoing learning systems together with cross-speciality 
supervision. Pathology departments should establish 
“AI governance committees” that unite medical 
specialists and bioinformatics professionals with legal 
experts and moral ethicists for checking new tools using 
standardised testing protocols and clinical effectiveness 
metrics (Gaffney & Mirza 2025). The necessary iterative 
refinement process requires performance dashboards 
which include indicators for model drift assessment and 
demographic sensitivity, and clinical impact evaluation 
(Panayides et al., 2020). To advance precision diagnostics 
as an everyday practice which handles ethical standards, 
pathologists must implement organized investments that 
create secure infrastructure and federated architectures 
and adaptive governance, together with XAI capabilities 
and cultural changes.



Pa
ge

 
10

9

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

Recommendation
The research findings produce multiple essential 
recommendations which direct the course of  clinical 
practice, together with research activities and policy 
development. Every health institution needs to adopt 
AI and genomic instruments into digital pathology 
systems to boost diagnostic accuracy while providing 
individualized care. The integration of  AI with genomic 
tools requires extensive training to build pathologists’ 
abilities for responsible interpretation of  AI-produced 
results together with genomic information. Policy officials 
and healthcare administrators need to create uniform 
guidelines that handle ethical problems, reveal algorithm 
operations, and guard patient information databases as the 
use of  federated learning and data collaborations increases.
The development of  explainable AI requires an active 
enhancement of  interdisciplinary relationships between 
computer scientists with molecular biologists, and 
clinicians to promote technological advances which 
remain clinically useful and patient-focused. Research 
funding for the evaluation of  AI diagnostic systems in 
different patient populations remains necessary to stop 
health equity gaps from forming. Embracing people-
centred, transparent, ethical methods represents the vital 
approach for achieving a complete medical revolution 
through AI genomics and molecular imaging.

CONCLUSION 
This systematic review shows that integrating artificial 
intelligence (AI), genomics, and molecular imaging 
improves diagnostic precision and personalization in 
pathology. AI enhances image interpretation and predictive 
accuracy; genomics deepens molecular classification; and 
molecular imaging visualizes in vivo processes. Together, 
these modalities accelerate workflows, reduce error, and 
strengthen treatment planning and outcomes.
Major findings indicate strong performance: deep-learning 
histopathology models frequently achieved AUCs of  0.90 
or higher, and multimodal systems that combine imaging 
with genomic features produced roughly a 15% gain in 
prognostic accuracy in comparative studies. Molecular 
imaging methods such as PET and SPECT demonstrated 
sensitivities exceeding 90% for detecting small lesions. 
In oncology stratification, integrating next-generation 
sequencing with histopathology achieved sensitivities 
above 95% in selected settings.
Clinical adoption remains constrained by data-format and 
interoperability gaps, privacy and bias risks, and limited 
explainability. Addressing these barriers requires robust 
data governance, investment in scalable compute and 
storage, and cross-disciplinary collaboration to develop 
transparent, validated models suitable for routine care.
Overall, the convergence of  AI, genomics, and molecular 
imaging marks a shift toward precision diagnostics. With 
explainable AI and ethical oversight, these integrated 
approaches can deliver faster, reliable decisions, establishing 
a more patient-centered paradigm for pathology and 
improving outcomes across healthcare systems.

REFERENCES
Abasher, M. M., Sidahmed, A. A., Abdelmoniem, A., 

& Mohamed, A. H. (2023). Breast cancer in elderly 
females. American Journal of  Medical Science and 
Innovation, 2(2), 16–25. https://doi.org/10.54536/
ajmsi.v2i2.1711

Ahmad, Z., Rahim, S., Zubair, M., & Abdul-Ghafar, 
J. (2021). Artificial intelligence (AI) in medicine, 
current applications and future role with special 
emphasis on its potential and promise in pathology: 
Present and future impact, obstacles including costs 
and acceptance among pathologists, practical and 
philosophical considerations. Diagnostic Pathology, 16, 
24. https://doi.org/10.1186/s13000-021-01085-4

Ahuja, S., & Zaheer, S. (2025). Advancements in 
pathology: Digital transformation, precision medicine, 
and beyond. Journal of  Pathology Informatics, 16, 100408. 
https://doi.org/10.1016/j.jpi.2024.100408

Ali, K., & Saqib, M. (2023). Risk prediction of  thalassemia 
using data mining classifiers. American Journal of  
Medical Science and Innovation, 2(2), 97–109. https://doi.
org/10.54536/ajmsi.v2i2.1979

Arvaniti, E., Berisha, V., Hagemann, V., et al. (2024). 
Explainable AI for computational pathology identifies 
model limitations and tissue biomarkers. arXiv preprint 
arXiv:2409.03080. https://arxiv.org/abs/2409.03080

Asif, A., Rajpoot, K., Graham, S., Snead, D., Minhas, F., 
& Rajpoot, N. (2023). Unleashing the potential of  
AI for pathology: challenges and recommendations. 
The Journal of  Pathology, 260(5), 564–577. https://doi.
org/10.1002/path.6168

Chartrand, G., Hager, G. D., & Avila, M. (2022). 
Explainable medical imaging AI needs human-
centered design: Guidelines and evidence from a 
systematic review. npj Digital Medicine, 5, 156. https://
doi.org/10.1038/s41746-022-00699-2

Cheng, J. Y., Abel, J. T., Balis, U. G. J., McClintock, D. 
S., & Pantanowitz, L. (2021). Challenges in the 
development, deployment, and regulation of  artificial 
intelligence in anatomic pathology. The American 
Journal of  Pathology, 191(10), 1684–1692. https://doi.
org/10.1016/j.ajpath.2020.10.018

Fahmy, A. M. (2024). High-resolution ultrasonography 
versus MRI in the diagnosis of  Achilles tendon lesions. 
American Journal of  Medical Science and Innovation, 3(2), 
52–58. https://doi.org/10.54536/ajmsi.v3i2.2801

He, W., Huang, W., Zhang, L., Wu, X., Zhang, S., & Zhang, 
B. (2024). Radiogenomics: Bridging the gap between 
imaging and genomics for precision oncology. 
MedComm, 5(9), e722. https://doi.org/10.1002/
mco2.722

Ikwuka, A. O., Musa, S., Udeh, F. C., Musa, A. A., & 
Chukwuezie, U. C. (2023). CRISPR-Cas9 genomic 
editing as an innovation in the management of  sickle 
cell disease: A systematic review. American Journal of  
Medical Science and Innovation, 2(2), 44–56. https://doi.
org/10.54536/ajmsi.v2i2.1760

Kermany, D. S., Goldbaum, M., Cai, W., Valentim, C. C. S., 



Pa
ge

 
11

0

https://journals.e-palli.com/home/index.php/ajmsi

Am. J. Med. Sci. Innov. 4(2) 103-110, 2025

Liang, H., Baxter, S. L., McKeown, A., Yang, G., Wu, 
X., Yan, F., Dong, J., Prasadha, M. K., Pei, J., Ting, 
M. Y. L., Zhu, J., Li, C., Hewett, S., Dong, J., Ziyar, 
I., Shi, A., … Zhang, K. (2018). Identifying medical 
diagnoses and treatable diseases by image-based deep 
learning. Cell, 172(5), 1122–1131.e9. https://doi.
org/10.1016/j.cell.2018.02.010

Kim, I., Kang, K., Song, Y., & Kim, T.-J. (2022). 
Application of  artificial intelligence in pathology: 
Trends and challenges. Diagnostics, 12(11), 2794. 
https://doi.org/10.3390/diagnostics12112794

Madabhushi, A., & Lee, G. (2016). Image analysis and 
machine learning in digital pathology: Challenges and 
opportunities. Medical Image Analysis, 33, 170–175. 
https://doi.org/10.1016/j.media.2016.06.037

Munari, E., Scarpa, A., Cima, L., Pozzi, M., Pagni, F., 
Vasuri, F., Marletta, S., Dei Tos, A. P., & Eccher, A. 
(2024). Cutting-edge technology and automation in 
the pathology laboratory: Review and perspectives. 
Virchows Archiv, 484, 555–566. https://doi.
org/10.1007/s00428-023-03655-6

Oala, L., Flach, P., & Ghalwash, M. (2022). Trust in 
artificial intelligence for medical diagnoses: A review. 
AI in Healthcare, 2, 100031. https://doi.org/10.1016/j.
aihc.2022.100031

Panayides, A. S., Amini, A., Filipovic, N. D., Sharma, 
A., Tsaftaris, S. A., Young, A., et al. (2020). AI in 
medical imaging informatics: Current challenges 
and future directions. IEEE Journal of  Biomedical and 
Health Informatics, 24(7), 1837–1846. https://doi.
org/10.1109/JBHI.2020.2986741

Prevedello, L. M., Halabi, S. S., Shih, G., Wu, C. C., Kohli, 
M. D., Chokshi, F. H., … & Flanders, A. E. (2019). 
Challenges related to artificial intelligence research in 
medical imaging and the importance of  image analysis 
competitions. Radiology: Artificial Intelligence, 1(1), 
e180031. https://doi.org/10.1148/ryai.20191800310

Rai, A. (2020). Explainable AI: From black box to glass 
box. Journal of  the Academy of  Marketing Science, 48(1), 
137–141. https://doi.org/10.1007/s11747-019-
00710-5

Rieke, N., Hancox, J., Li, W., Milletari, F., Roth, H. R., 
Albarqouni, S., … & Summers, R. M. (2020). The 
future of  digital health with federated learning. npj 
Digital Medicine, 3, 119. https://doi.org/10.1038/
s41746-020-00323-1

Rowe, S. (2021). Artificial intelligence in molecular 
imaging: At the crossroads of  revolutions in medical 
diagnosis. Annals of  Translational Medicine, 9(10), 826. 
https://doi.org/10.21037/atm-20-6191

Salo, I., Nordlund, L., Eklund, L., Ho, J., Soini, M., 
Kumar, D., Yeong, J., Guan, F., & Metsälä, E. (2024). 
Advancements and applications of  AI technologies 
in pathology: A scoping review. Computer Methods 
in Biomechanics and Biomedical Engineering: Imaging & 
Visualization, 12(1). https://doi.org/10.1080/216811
63.2024.2396595

Saxena, S., Jena, B., Gupta, N., Das, S., Sarmah, D., 
Bhattacharya, P., Nath, T., Paul, S., Fouda, M. M., 
Kalra, M., Saba, L., Pareek, G., & Suri, J. S. (2022). Role 
of  artificial intelligence in radiogenomics for cancers 
in the era of  precision medicine. Cancers, 14(12), 2860. 
https://doi.org/10.3390/cancers14122860

Seyhan, A. A., & Carini, C. (2019). Are innovation and 
new technologies in precision medicine paving a new 
era in patient-centric care? Journal of  Translational 
Medicine, 17, 114. https://doi.org/10.1186/s12967-
019-1901-0

Shafi, S., & Parwani, A. V. (2023). Artificial intelligence 
in diagnostic pathology. Diagnostic Pathology, 18, 109. 
https://doi.org/10.1186/s13000-023-01375-z

Sussman, L., Garcia-Robledo, J. E., Ordóñez-Reyes, C., 
Forero, Y., Mosquera, A. F., Ruíz-Patiño, A., Chamorro, 
D. F., & Cardona, A. F. (2022). Integration of  artificial 
intelligence and precision oncology in Latin America. 
Frontiers in Medical Technology, 4, 1007822. https://doi.
org/10.3389/fmedt.2022.1007822

Xu, J., Glicksberg, B. S., Su, C., Walker, P., Bian, J., & 
Wang, F. (2020). Federated learning for healthcare 
informatics. Journal of  Healthcare Informatics Research, 
5(1), 1–19. https://doi.org/10.1007/s41666-020-
00082-4


