Matches in SemOpenAlex for { <https://semopenalex.org/work/W4367000365> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4367000365 abstract "Deep learning has made great strides in medical imaging, enabled by hardware advances in GPUs. One major constraint for the development of new models has been the saturation of GPU memory resources during training. This is especially true in computational pathology, where images regularly contain more than 1 billion pixels. These pathological images are traditionally divided into small patches to enable deep learning due to hardware limitations. In this work, we explore whether the shared GPU/CPU memory architecture on the M1 Ultra systems-on-a-chip (SoCs) recently released by Apple, Inc. may provide a solution. These affordable systems (less than $5000) provide access to 128 GB of unified memory (Mac Studio with M1 Ultra SoC). As a proof of concept for gigapixel deep learning, we identified tissue from background on gigapixel areas from whole slide images (WSIs). The model was a modified U-Net (4492 parameters) leveraging large kernels and high stride. The M1 Ultra SoC was able to train the model directly on gigapixel images (16000$times$64000 pixels, 1.024 billion pixels) with a batch size of 1 using over 100 GB of unified memory for the process at an average speed of 1 minute and 21 seconds per batch with Tensorflow 2/Keras. As expected, the model converged with a high Dice score of 0.989 $pm$ 0.005. Training up until this point took 111 hours and 24 minutes over 4940 steps. Other high RAM GPUs like the NVIDIA A100 (largest commercially accessible at 80 GB, $sim$$15000) are not yet widely available (in preview for select regions on Amazon Web Services at $40.96/hour as a group of 8). This study is a promising step towards WSI-wise end-to-end deep learning with prevalent network architectures." @default.
- W4367000365 created "2023-04-27" @default.
- W4367000365 creator A5010406734 @default.
- W4367000365 creator A5012851448 @default.
- W4367000365 creator A5037133367 @default.
- W4367000365 creator A5048972725 @default.
- W4367000365 creator A5058681839 @default.
- W4367000365 creator A5067191302 @default.
- W4367000365 creator A5067212450 @default.
- W4367000365 creator A5073353188 @default.
- W4367000365 creator A5075735203 @default.
- W4367000365 creator A5089262488 @default.
- W4367000365 creator A5089717791 @default.
- W4367000365 date "2023-04-24" @default.
- W4367000365 modified "2023-09-28" @default.
- W4367000365 title "Exploring shared memory architectures for end-to-end gigapixel deep learning" @default.
- W4367000365 doi "https://doi.org/10.48550/arxiv.2304.12149" @default.
- W4367000365 hasPublicationYear "2023" @default.
- W4367000365 type Work @default.
- W4367000365 citedByCount "0" @default.
- W4367000365 crossrefType "posted-content" @default.
- W4367000365 hasAuthorship W4367000365A5010406734 @default.
- W4367000365 hasAuthorship W4367000365A5012851448 @default.
- W4367000365 hasAuthorship W4367000365A5037133367 @default.
- W4367000365 hasAuthorship W4367000365A5048972725 @default.
- W4367000365 hasAuthorship W4367000365A5058681839 @default.
- W4367000365 hasAuthorship W4367000365A5067191302 @default.
- W4367000365 hasAuthorship W4367000365A5067212450 @default.
- W4367000365 hasAuthorship W4367000365A5073353188 @default.
- W4367000365 hasAuthorship W4367000365A5075735203 @default.
- W4367000365 hasAuthorship W4367000365A5089262488 @default.
- W4367000365 hasAuthorship W4367000365A5089717791 @default.
- W4367000365 hasBestOaLocation W43670003651 @default.
- W4367000365 hasConcept C108583219 @default.
- W4367000365 hasConcept C111919701 @default.
- W4367000365 hasConcept C149635348 @default.
- W4367000365 hasConcept C154945302 @default.
- W4367000365 hasConcept C160633673 @default.
- W4367000365 hasConcept C173608175 @default.
- W4367000365 hasConcept C41008148 @default.
- W4367000365 hasConcept C74296488 @default.
- W4367000365 hasConcept C9390403 @default.
- W4367000365 hasConcept C98045186 @default.
- W4367000365 hasConceptScore W4367000365C108583219 @default.
- W4367000365 hasConceptScore W4367000365C111919701 @default.
- W4367000365 hasConceptScore W4367000365C149635348 @default.
- W4367000365 hasConceptScore W4367000365C154945302 @default.
- W4367000365 hasConceptScore W4367000365C160633673 @default.
- W4367000365 hasConceptScore W4367000365C173608175 @default.
- W4367000365 hasConceptScore W4367000365C41008148 @default.
- W4367000365 hasConceptScore W4367000365C74296488 @default.
- W4367000365 hasConceptScore W4367000365C9390403 @default.
- W4367000365 hasConceptScore W4367000365C98045186 @default.
- W4367000365 hasLocation W43670003651 @default.
- W4367000365 hasOpenAccess W4367000365 @default.
- W4367000365 hasPrimaryLocation W43670003651 @default.
- W4367000365 hasRelatedWork W1604898313 @default.
- W4367000365 hasRelatedWork W2117014006 @default.
- W4367000365 hasRelatedWork W2362713694 @default.
- W4367000365 hasRelatedWork W2570873234 @default.
- W4367000365 hasRelatedWork W2731899572 @default.
- W4367000365 hasRelatedWork W2794115703 @default.
- W4367000365 hasRelatedWork W2939353110 @default.
- W4367000365 hasRelatedWork W3009238340 @default.
- W4367000365 hasRelatedWork W3215138031 @default.
- W4367000365 hasRelatedWork W4308233823 @default.
- W4367000365 isParatext "false" @default.
- W4367000365 isRetracted "false" @default.
- W4367000365 workType "article" @default.