Federated Learning with Research Prototypes: Application to Multi-Center MRI-based Detection of Prostate Cancer with Diverse Histopathology.

RATIONALE AND OBJECTIVES: Early prostate cancer detection and staging from MRI is extremely challenging for both radiologists and deep learning algorithms, but the potential to learn from large and diverse datasets remains a promising avenue to increase their performance within and across institutions. To enable this for prototype-stage algorithms, where the majority of existing research remains, we introduce a flexible federated learning framework for cross-site training, validation, and evaluation of custom deep learning prostate cancer detection algorithms. MATERIALS AND METHODS: We introduce an abstraction of prostate cancer groundtruth that represents diverse annotation and histopathology data. We maximize use of this groundtruth if and when they are available using UCNet, a custom 3D UNet that enables simultaneous supervision of pixel-wise, region-wise, and gland-wise classification. We leverage these modules to perform cross-site federated training using 1400+ heterogeneous multi-parameteric prostate MRI exams from two University hospitals. RESULTS: We observe a positive result, with significant improvements in cross-site generalization performance with negligible intra-site performance degradation for both lesion segmentation and per-lesion binary classification of clinically-significant prostate cancer. Cross-site lesion segmentation performance intersection-over-union (IoU) improved by 100%, while cross-site lesion classification performance overall accuracy improved by 9.5-14.8%, depending on the optimal checkpoint selected by each site. CONCLUSION: Federated learning can improve the generalization performance of prostate cancer detection models across institutions while protecting patient health information and institution-specific code and data. However, even more data and participating institutions are likely required to improve the absolute performance of prostate cancer classification models. To enable adoption of federated learning with limited re-engineering of federated components, we open-source our FLtools system at https://federated.ucsf.edu, including examples that can be easily adapted to other medical imaging deep learning projects.

Detecting, extracting, and mapping of inland surface water using Landsat 8 Operational Land Imager: A case study of Pune district, India.

Background: Recent developments in optical satellite remote sensing have led to a new era in the detection of surface water with its changing dynamics. This study presents the creation of surface water inventory for a part of Pune district (an administrative area), in India using the Landsat 8 Operational Land Imager (OLI) and a multi spectral water indices method. Methods: A total of 13 Landsat 8 OLI cloud free images were analyzed for surface water detection. Modified Normalized Difference Water Index (MNDWI) spectral index method was employed to enhance the water pixels in the image. Water and non-water areas in the map were discriminated using the threshold slicing method with a trial and error approach. The accuracy analysis based on kappa coefficient and percentage of the correctly classified pixels was presented by comparing MNDWI maps with corresponding Joint Research Centre (JRC) Global Surface Water Explorer (GSWE) images. The changes in the surface area of eight freshwater reservoirs within the study area (Bhama Askhed, Bhatghar, Chaskaman, Khadakwasala, Mulashi, Panshet, Shivrata, and Varasgaon) for the year 2016 were analyzed and compared to GSWE time series water databases for accuracy assessment. The annual water occurrence map with percentage water occurrence on a yearly basis was also prepared. Results: The kappa coefficient agreement between MNDWI images and GSWE images is in the range of 0.56 to 0.96 with an average agreement of 0.82 indicating a strong level of agreement. Conclusions: MNDWI is easy to implement and is a sufficiently accurate method to separate water bodies from satellite images. The accuracy of the result depends on the clarity of image and selection of an optimum threshold method. The resulting accuracy and performance of the proposed algorithm will improve with implementation of automatic threshold selection methods and comparative studies for other spectral indices methods.

Federated learning improves site performance in multicenter deep learning without data sharing.

OBJECTIVE: To demonstrate enabling multi-institutional training without centralizing or sharing the underlying physical data via federated learning (FL). MATERIALS AND METHODS: Deep learning models were trained at each participating institution using local clinical data, and an additional model was trained using FL across all of the institutions. RESULTS: We found that the FL model exhibited superior performance and generalizability to the models trained at single institutions, with an overall performance level that was significantly better than that of any of the institutional models alone when evaluated on held-out test sets from each institution and an outside challenge dataset. DISCUSSION: The power of FL was successfully demonstrated across 3 academic institutions while avoiding the privacy risk associated with the transfer and pooling of patient data. CONCLUSION: Federated learning is an effective methodology that merits further study to enable accelerated development of models across institutions, enabling greater generalizability in clinical use.