Whole Slide Imaging in Genitourinary Pathology:
A Cloud-Based Digitisation Workflow for Resource-Limited Settings

By Aisha Memon¹, Talat Zehra², Sarosh Moeen¹, Mahin Shams³, Anil Parwani⁴, Saad Nadeem⁵

Affiliations

1. Department of Pathology and Laboratory Medicine, The Aga Khan University Hospital, Karachi, Pakistan
2. Department of Pathology, Jinnah Sindh Medical University, Karachi. Pakistan
3. Department of Pathology, United Medical and Dental College, Karachi, Pakistan
4. Department of Pathology, The Ohio State University, Columbus, USA
5. Department of Medical Physics, Memorial Sloan Kettering Cancer Centre, New York, USA

doi: 10.29271/jcpsp.2025.12.1579

ABSTRACT
Objective: To digitise genitourinary glass slides and use them for research, education, and knowledge sharing.
Study Design: Observational study.
Place and Duration of the Study: Department of Pathology and Laboratory Medicine, The Aga Khan University Hospital, Karachi, Pakistan, from May 2025 to June 2025.
Methodology: This study used 31 renal biopsy cores with special stains and 21 prostate tissue cores with haematoxylin and eosin stains. Videos at 10x magnification were recorded using a microscope-mounted camera and uploaded to the DeepLIIF open-source cloud platform to generate whole-slide images (WSIs). Three pathologists independently reviewed the images, and inter-observer agreement was assessed using the Kappa statistic.
Results: Pathologist A showed 100% agreement with the reference diagnosis across all renal stains. Pathologist B demonstrated concordance of 77% for haemoxylin and eosin (H&E) and 74% for periodic Acid-Schiff (PAS), Jones methenamine silver (JMS), and trichrome stains. Pathologist C reported 100% agreement for H&E and PAS, 96% for trichrome, and 74% for JMS. In the prostate biopsies, among 21 cases (6 malignant and 15 benign), the diagnosis regarding malignancy and perineural invasion was compared to the reference diagnosis.
Conclusion: WSIs can be viewed through shareable links, even on smartphones. This helps in getting consensus on improved diagnosis, especially in resource-limited settings.

Key Words: Digitisation, Genitourinary pathology, Image interpretation, Telepathology.

INTRODUCTION

Digitising histology slides is revolutionising workflow in histopathology. It is gradually becoming essential in modern patho-logy practice, particularly for remote consultations (telepatho- logy), education, and artificial intelligence (AI)-assisted diag- nostics.^1,2 These challenges are more apparent in genitouri- nary pathology, where accurate diagnosis—particularly of renal biopsy—requires high-resolution imaging for detailed visuali- sation of glomerular, tubular, interstitial, and vascular compartments,  as  well  as  access  to  multiple  histochemical  stains.

Although renal biopsy remains the gold standard for diagnosing kidney diseases, the rising incidence of renal diseases and the shortage of renal pathologists have created a pressing need for new diagnostic models. AI and digital pathology offer promising solutions for improving accuracy in detection, classification, and outcome prediction.³

Whole-slide imaging (WSI), also known as virtual microscopy, involves scanning glass slides to produce digital images. It enables pathologists working in rural or remote areas to seek expert opinions from subspecialists in urban centres. This technology facilitates the rapid sharing of knowledge, allowing rare cases or region-specific histopathological patterns to be spread across national and international platforms. The digitised slides also serve as a basis for developing, testing, and validating AI-based computational algorithms on diverse cohorts from various demographics and different settings.^4,5 However, current digitisation workflows are expensive to establish and maintain in the long run.^2,4 Commercial WSI platforms require digital multi-slide scanners, dedicated space for slide scanning, high-resolution cameras, cloud or on-premises data storage facilities, specialised IT, and technical personnel.^1,2 Nearly two-thirds of the global population resides in low- and middle-income countries, where the number of trained pathologists is rapidly declining. These resources are beyond the reach of most hospitals in these countries, so the use of this techno- logy  is  limited.⁶

As a result, a gap in digital pathology access persists, impacting patient outcomes and limiting training opportunities for pathology residents and technicians in resource-constrained countries.² A user-friendly, cloud-based tool, HistoCloud, has been developed for whole-slide image segmentation that allows pathologists to segment structures, including glomeruli and interstitial fibrosis, without the need for programming expertise. This  can  facilitate  a  wider  adoption  in  clinical  settings.⁷

Previous research studies have used histopathological images and applied an AI-based deep learning model.^8-10 In this project, previously diagnosed single cores of predominantly renal and a few prostatic cores were used. This study aimed to develop and evaluate a cost-effective, cloud-based slide digitisation work-flow using low-resolution video microscopy and open-source tools. The generated images were stitched into high-resolution whole- slide images (WSIs). This approach aimed to enhance access to digital pathology and AI applications in resource-limited settings, particularly for the evaluation of renal and prostatic  tissues.

METHODOLOGY

This study included a total of 31 previously diagnosed renal core biopsies (non-neoplastic) and 21 prostatic core biopsies, encompassing both benign and malignant lesions. Ethical approval was obtained from the Ethics Review Committee of the Agha Khan University Hospital, Karachi, Pakistan (ERC No. 2025-11452- 34443) prior to data collection.

Glass slides were examined manually using standard light microscopy. During slide review at 10x magnification, video recordings were captured and subsequently processed using the DeepLIIF Cloud platform—an open-source, deep learning-based system for generating high-quality WSIs through automated stitching. In the renal biopsy evaluation, a total of 124 WSIs were prepared from 31 cases. The resulting WSIs were stitched together and subsequently reviewed by three independent pathologists (designated as Pathologists A, B, and C). For renal biopsies, the original diagnoses were shared with the pathologists prior to review. Pathologists were asked to assess the images for research, education, knowledge sharing and telepathology purposes and for individual cases and special stains. For prostatic biopsies, the pathologists were blinded to the original diagnosis and were asked to assess each core for benign vs. malignant features and for the presence of perineural invasion. In cases of prostate, only haematoxylin and eosin-stained slides were used. As all the biopsy cases were previously diagnosed, the original diagnosis was considered the gold standard and used for comparison.

The renal biopsy specimens were single-core samples and included a range of pathological diagnoses, including minimal change disease (MCD), focal segmental glomerulosclerosis (FSGS), membranous glomerulonephritis (MGN), membranoproliferative glomerulonephritis (MPGN), mesangio-proliferative glomerulonephritis, lupus nephritis, diabetic nephropathy, renal amyloidosis, and renal allograft biopsies. Four stains were utilised for renal biopsies: haematoxylin and eosin (H&E), periodic acid-Schiff (PAS), Jones methenamine silver (JMS), and Masson’s  trichrome  stain.

The inclusion criteria included single-core renal or prostatic biopsies, digitised slides with high-quality staining across diverse renal pathologies, and the use of at least one of the four stains mentioned above for renal cases. Slides with poor staining quality were excluded from the study. The annotation tool was used manually by a histopathologist for marking glomeruli in renal biopsies and perineural invasion in prostatic core  biopsies,  as  shown  in  Figure  1-3.

Figure 1: (A) Renal biopsy whole-slide image at low magnification. (B, C) The same core at 10x and 20x magnifications, respectively. Glomeruli were manually annotated using an annotation tool.

Figure 2: (A-D) Renal biopsy of membranous glomerulonephritis showing H&E and special stains at 5x magnification.

Figure 3: (A, B) Prostatic core showing whole-slide image at low power. The same core at 10x shows prostatic adenocarcinoma invading a nerve, annotated in orange colour using the annotation tool. (Inset: Navigation map is also present in both images)

Statistical analysis was performed using STATA 18. Inter- observer agreement among the pathologists was assessed using the Kappa statistic, and a p-value of <0.05 was consi- dered significant.

RESULTS

The diagnostic agreement among the three pathologists varied across different stains. Pathologist A demonstrated complete concordance, showing 100% agreement with the reference diagnosis across all stains. Pathologist B showed 77% concordance for H&E and 74% for PAS, JMS, and trichrome stains. In comparison, Pathologist C achieved 100% agreement for H&E and PAS, 96% for trichrome, and 74% for JMS. In the prostate biopsy cohort, among 21 cases (6 malignant and 15 benign), WSIs were assessed by the same panel of pathologists in a blinded manner. Diagnostic conclusions regarding malignancy and perineural invasion were compared to the original diagnosis (gold standard). The specific accuracy rates for each pathologist are summarised in Table I. The p-values were calculated  using  the  kappa  statistic.

DISCUSSION

Digitisation of histopathological slides, particularly in renal and genitourinary pathology, offers several advantages, including efficient case archiving, rapid retrieval for second opinions, and enhanced comparison of stains through multiplex viewing without the need for physical slide exchange.^11,12 WSI supports AI-assisted image analysis for glomerular identification, sclerosis assessment, and automated quantification, improving both accuracy and workflow consistency.¹³

Studies have shown that WSI achieves diagnostic accuracy comparable to traditional light microscopy, with high concordance in large-scale validation trials. Mukhopadhyay et al. demonstrated that WSI was comparable to conventional microscopy in diagnosing complex surgical pathology.¹³ Platforms such as OpenSlide and PathPresenter have facilitated in the cost-effective adoption of WSI.¹⁴ In addition to allowing access to digital pathology, these tools also support education and teleconsultation, empowering local clinicians with diagnostic support and ongoing training opportunities. Cloud-based digital pathology systems reduce dependence on local IT infrastructure and enable flexible access, which is especially beneficial in low- and middle-income countries.¹⁵ Collaborative initiatives such as the PANDA challenge have demonstrated how cloud-hosted repositories and AI-assisted models can foster global research and consistency in genitourinary diagnosis.¹⁶

Despite these advantages, implementing WSI in resource-limited settings remains challenging due to factors such as poor internet connectivity, digital literacy gaps, and varying regulatory standards.¹⁷ However, innovations in low-cost scanners, open-source AI tools, and hybrid edge-cloud systems such as Aiosyn highlight a promising future for the widespread adoption of cloud-based digital pathology.^18,19

While the digital quality in this study was generally good, some limitations were noted. Renal biopsies often require serial sectioning and multi-level review, as many diseases, such as FSGS, exhibit focal lesions that may be missed on a single level. This workflow currently digitises only a single histological level per slide, restricting its utility for diseases that require multi-level assessment. Moreover, artefacts such as tissue folding, uneven sectioning, and inconsistent staining may compromise image clarity and diagnostic reliability in some cases. Conti- nuous focus was required during video recording, and acquisition was generally feasible at 10x magnification; blurring was more common at 20x due to focusing issues. Files larger than 1.5 GB could not be uploaded, and the workflow did not include immunofluorescence digitisation. Despite these factors, the digitised images enabled detailed review of key renal compartments, including glomerular, tubular, interstitial, and vascular structures.

Although the study was small, its results were really promising. The pathologists were able to view the whole case, along with special stains, in a single go. Most of the non-neoplastic renal pathologies could usually be diagnosed at 20x, so blurring issues at 40x were not very challenging in most cases. WSIs could be viewed through shareable links, even on smartphones. Therefore, in regions where technology limitations persist, such solutions can be particularly valuable, especially for renal patho- logies that require specialist services in most cases. In deve- loping countries, the number of specialised genitourinary patho- logists is very limited.^20,21 Most renal biopsies are performed in larger or specialised setups, so shareable links enable patho- logists worldwide access and review various renal pathologies in real time, providing equal learning opportunities. Similarly, small or focal prostatic malignancies may require the expertise of a genitourinary pathologist. This system allows pathologist to seek expert  opinions  from  anywhere  in  the  world.^22,23

CONCLUSION

Digital methods are not a replacement for scanners; however, such tools can be highly valuable in resource-limited settings of low- and middle-income countries. Small steps such as this help extend the benefits of digitisation and computation tools to those who may not have access to these novel tools, particularly in  regions  where  technology  glitches  are  common.

ETHICAL  APPROVAL:
Ethical approval was obtained from the Ethics Review Committee of the Agha Khan University Hospital, Karachi, Pakistan (ERC No. 2025-11452-34443). This study was conducted in accordance with  the  ethical  standards.

PATIENTS’  CONSENT:
This study used previously diagnosed and archived biopsy specimens. As the research involved anonymised retrospective data with no direct patient identifiers, patient consent was not needed.

COMPETING  INTEREST:
The  authors  declared  no  conflict  of  interest.

AUTHORS’  CONTRIBUTION:
AM: Supervised the project, conceived and analysed the data, and  contributed  to  manuscript  writing.
TZ: Collected data, digitised slides, and wrote the manuscript.
SM:  Collected  data  and  compiled  results.
MS: Contributed to manuscript writing and compiled the results.
AP: Critically analysed the project and contributed to the results compilation.
SN: Stitched slides into whole-slide images and performed critical analyses.
All authors approved the final version of the manuscript to be published.

REFERENCES

1. Zehra T, Marino J, Wang W, Frantsuzov G, Nadeem S. Rethinking histology slide digitisation workflows for low-resource settings. In book: Medical Image Computing and Computer Assisted Intervention – MICCAI 2024. p.427-36. doi: 10.1007/978-3-031-72083-3_40.
2. Patel A, Balis UJG, Cheng J, Li Z, Lujan G, McClintock DS, et al. Contemporary whole slide imaging devices and their applications within the modern pathology department: A selected hardware review. J Pathol Inform 2021; 12:50. doi: 10.4103/jpi.jpi_66_21.
3. Feng C, Liu F. Artificial intelligence in renal pathology: Current status and future. Biomol Biomed 2023; 23(2): 225-34. doi: 10.17305/bjbms.2022.8318.
4. Farahani N, Parwani AV, Pantanowitz L. Whole slide imaging in pathology: Advantages, limitations, and emerging perspectives. Pathol Lab Med Int 2015; 7:23-33. doi: 10. 2147/PLMI.S59826.
5. Niazi MKK, Parwani AV, Gurcan MN. Digital pathology and artificial intelligence. Lancet Oncol 2019; 20(5):e253-61. doi: 10.1016/S1470-2045(19)30154-8.
6. Zehra T, Ahmed Z, Shams M. Current practice of histo-pathology in Pakistan: difficulties, challenges, and solutions. J Pak Med Assoc 2023; 73(7):1488-90. doi: 10.47391/JPMA. 7262.
7. Lutnick B, Manthey D, Becker JU, Ginley B, Moos K, Zuckerman JE, et al. A user-friendly tool for cloud-based whole slide image segmentation with examples from renal histopathology. Commun Med (Lond) 2022; 2:105. doi: 10.1038/s43856-022-00138-z.
8. Rahaman MM, Millar EKA, Meijering E. Generalized deep learning for histopathology image classification using super-vised contrastive learning. J Adv Res 2025; 75:389-404. doi: 10.1016/j.jare.2024.11.013.
9. Unger M, Kather JN. Deep learning in cancer genomics and histopathology. Genome Med 2024; 16(1):44. doi: 10.1186/ s13073-024-01315-6.
10. Wu T, Wang Q, Li Y, Xu K, Chen D, Xu H, et al. Diagnostic accuracy of deep learning in detection and prognostication of renal cell carcinoma: A systematic review and meta-analysis. Front Med (Lausanne) 2024; 11:1447057. doi: 10.3389/fmed.2024.1447057.
11. L’Imperio V, Brambilla V, Cazzaniga G, Ferrario F, Nebuloni M, Pagni F. Digital pathology for the routine diagnosis of renal diseases: A standard model. J Nephrol 2021; 34(3): 681-8. doi: 10.1007/s40620-020-00847-1.
12. Neri F, Eccher A, Rigotti P, Girolami I, Zaza G, Gambaro G, et al. Advantages of using a web-based digital platform for kidney preimplantation biopsies. J Pathol Inform 2021; 12(1):41. doi: 10.4103/jpi.jpi_45_21.
13. Mukhopadhyay S, Feldman MD, Abels E, Ashfaq R, Beltaifa S, Cacciabeve NG, et al. Whole slide imaging versus microscopy for primary diagnosis in surgical pathology: A multicenter blinded randomized noninferiority study of 1992 cases (Pivotal Study). Am J Surg Pathol 2018; 42(1): 39-52. doi: 10.1097/PAS.0000000000000948.
14. Brady AP, Lim-Dunham JE, Kwon R, Mollura DJ. Open-source software tools for digital pathology: A review. J Pathol Inform 2020; 11:22. doi: 10.4103/jpi.jpi_6_20.
15. Hanna MG, Reuter VE, Ardon O, Kim D, Sirintrapun SJ, Schuffler PJ, et al. Integrating digital pathology into clinical practice. Mod Pathol 2019; 32(9):1205-14. doi: 10.1038/ s41379-019-0322-6.
16. Bulten W, Kartasalo K, Chen PC, Strom P, Pinckaers H, Nagpal K, et al. Artificial intelligence for diagnosis and Gleason grading of prostate cancer: The PANDA challenge. Nat Med 2022; 28(1):154-63. doi: 10.1038/s41591-021-01620-2.
17. Mremi A, Achola C, Mbwambo D, Magorosa E, Legason ID, Vavoulis D, et al. Diagnostic validation of a portable whole slide imaging scanner for lymphoma diagnosis in resource-constrained setting: A cross-sectional study. J Pathol Inform 2023; 14:100188. doi: 10.1016/j.jpi.2023.100188.
18. Barisoni L, Nast CC, Jennette JC, Hodgin JB, Herzenberg AM, Lemley KV, et al. Digital pathology and computational image analysis in nephropathology. Nat Rev Nephrol 2020; 16(10):610-24. doi: 10.1038/s41581-020-0319-2.
19. Ginley B, Jen K-Y, Han SS, Rodrigues L, Mankad S, Barisoni L, et al. Computational segmentation and classification of diabetic glomerulosclerosis. Kidney Int Rep 2021; 6(1): 40-9. doi: 10.1016/j.ekir.2020.10.003.
20. Bijol V, Farag YM, Harris DC, Levin A, Roy-Chaudhury P, Singh AK, et al. Renal pathology practice globally: Identi- fying needs and meeting the challenge. Kidney Int 2019; 96(2):258-61. doi: 10.1016/j.kint.2018.10.040.
21. Sharif MU, Elsayed ME, Stack AG. The global nephrology workforce: emerging threats and potential solutions. Clin Kidney J 2016; 9(1):11-22. doi: 10.1093/ckj/sfv111.
22. Shepherd DJ, Gordetsky JB. Digital pathology allows for global second opinions for urologic malignancies. Curr Urol Rep 2025; 26(1):1-7. doi: 10.1007/s11934-025-01255-7.
23. Tian YQ, Yang JC, Hu JJ, Ding R, Ye DW, Shang JW. Trends and risk factors of global incidence, mortality, and disability of genitourinary cancers from 1990 to 2019: Systematic analysis for the Global Burden of Disease Study 2019. Front Public Health 2023; 11:1119374. doi: 10.3389/fpubh.2023. 1119374.