Outcome prediction of prostate cancer patients on active surveillance using weakly supervised deep learning
(2025) SPIE Medical Imaging 2025 13413.- Abstract
- To avoid over-treatment of prostate cancer patients following screening for elevated prostate-specific antigen (PSA) levels, keeping patients on active surveillance has been suggested as an alternative to radical treatment. This means reoccurring visits for patients with low-grade cancer to monitor progression. The Prostate Cancer Research International Active Surveillance (PRIAS) study was initiated to research the benefits of active surveillance. Usually, the presence and grade of cancer are determined with needle-core biopsies of the prostate. However, the Gleason grading scale has demonstrated intra- and inter-observer variability. This is a limiting factor for novel automated Gleason grading algorithms. To address this, and... (More)
- To avoid over-treatment of prostate cancer patients following screening for elevated prostate-specific antigen (PSA) levels, keeping patients on active surveillance has been suggested as an alternative to radical treatment. This means reoccurring visits for patients with low-grade cancer to monitor progression. The Prostate Cancer Research International Active Surveillance (PRIAS) study was initiated to research the benefits of active surveillance. Usually, the presence and grade of cancer are determined with needle-core biopsies of the prostate. However, the Gleason grading scale has demonstrated intra- and inter-observer variability. This is a limiting factor for novel automated Gleason grading algorithms. To address this, and simultaneously utilizing a cohort collected within the PRIAS program, we have developed a deep learning-based framework for outcome prediction of patients on active surveillance. Our framework does not use explicit Gleason grades and consists of a pre-trained feature extractor, a feature selector, and an attention-based outcome predictor. We evaluate three feature extractors: the foundation model UNI, an ImageNet pre-trained model, and our Gleason grading network. Using UNI as the feature extractor outperformed the other models, with an average area under the receiver operator characteristic curve (AUC) of 0.996 (95% CI: 0.996 – 0.996). To our knowledge, this is the first end-to-end deep learning-based model for patient-level outcome predictions of prostate cancer patients on active surveillance. We believe that our algorithm could assist the pathologists and facilitate the implementation of prostate cancer screening programs, however, more work is needed in terms of validation and generalization. Our code for training and evaluating our models is publicly available*. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/0a7edd4f-b4a2-46a7-8d4c-3ee8bd12d841
- author
- Winzell, Filip
LU
; Arvidsson, Ida
LU
; Åström, Kalle LU
; Overgaard, Niels Christian LU ; Marginean, Felicia-Elena LU
; Simoulis, Athanasios LU
; Bjartell, Anders LU ; Krzyzanowska, Agnieszka LU and Heyden, Anders LU
- organization
-
- LTH Profile Area: Engineering Health
- Computer Vision and Machine Learning (research group)
- LU Profile Area: Natural and Artificial Cognition
- ELLIIT: the Linköping-Lund initiative on IT and mobile communication
- LU Profile Area: Proactive Ageing
- LTH Profile Area: AI and Digitalization
- eSSENCE: The e-Science Collaboration
- LU Profile Area: Nature-based future solutions
- LU Profile Area: Light and Materials
- Stroke Imaging Research group (research group)
- Mathematical Imaging Group (research group)
- Engineering Mathematics (M.Sc.Eng.)
- Partial differential equations (research group)
- LUCC: Lund University Cancer Centre
- Urological cancer, Malmö (research group)
- EpiHealth: Epidemiology for Health
- Division of Translational Cancer Research
- publishing date
- 2025-04-10
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- Medical Imaging 2025 : Digital and Computational Pathology - Digital and Computational Pathology
- editor
- Tomaszewski, John E. and Ward, Aaron D.
- volume
- 13413
- publisher
- SPIE
- conference name
- SPIE Medical Imaging 2025
- conference location
- San Diego, United States
- conference dates
- 2025-02-16 - 2025-02-20
- external identifiers
-
- scopus:105004791968
- DOI
- 10.1117/12.3046323
- language
- English
- LU publication?
- yes
- id
- 0a7edd4f-b4a2-46a7-8d4c-3ee8bd12d841
- date added to LUP
- 2025-04-11 08:52:23
- date last changed
- 2025-08-13 09:52:29
@inproceedings{0a7edd4f-b4a2-46a7-8d4c-3ee8bd12d841, abstract = {{To avoid over-treatment of prostate cancer patients following screening for elevated prostate-specific antigen (PSA) levels, keeping patients on active surveillance has been suggested as an alternative to radical treatment. This means reoccurring visits for patients with low-grade cancer to monitor progression. The Prostate Cancer Research International Active Surveillance (PRIAS) study was initiated to research the benefits of active surveillance. Usually, the presence and grade of cancer are determined with needle-core biopsies of the prostate. However, the Gleason grading scale has demonstrated intra- and inter-observer variability. This is a limiting factor for novel automated Gleason grading algorithms. To address this, and simultaneously utilizing a cohort collected within the PRIAS program, we have developed a deep learning-based framework for outcome prediction of patients on active surveillance. Our framework does not use explicit Gleason grades and consists of a pre-trained feature extractor, a feature selector, and an attention-based outcome predictor. We evaluate three feature extractors: the foundation model UNI, an ImageNet pre-trained model, and our Gleason grading network. Using UNI as the feature extractor outperformed the other models, with an average area under the receiver operator characteristic curve (AUC) of 0.996 (95% CI: 0.996 – 0.996). To our knowledge, this is the first end-to-end deep learning-based model for patient-level outcome predictions of prostate cancer patients on active surveillance. We believe that our algorithm could assist the pathologists and facilitate the implementation of prostate cancer screening programs, however, more work is needed in terms of validation and generalization. Our code for training and evaluating our models is publicly available*.}}, author = {{Winzell, Filip and Arvidsson, Ida and Åström, Kalle and Overgaard, Niels Christian and Marginean, Felicia-Elena and Simoulis, Athanasios and Bjartell, Anders and Krzyzanowska, Agnieszka and Heyden, Anders}}, booktitle = {{Medical Imaging 2025 : Digital and Computational Pathology}}, editor = {{Tomaszewski, John E. and Ward, Aaron D.}}, language = {{eng}}, month = {{04}}, publisher = {{SPIE}}, title = {{Outcome prediction of prostate cancer patients on active surveillance using weakly supervised deep learning}}, url = {{http://dx.doi.org/10.1117/12.3046323}}, doi = {{10.1117/12.3046323}}, volume = {{13413}}, year = {{2025}}, }