mlsanityData to 7 Oct 2026

PocketNet: A Smaller Neural Network for Medical Image Analysis

arXiv 2104.10745 ↗First posted 21 Apr 2021eess.IV cs.CV cs.LGcited by 9

Not on the map. Why?

Abstract (arXiv)

Medical imaging deep learning models are often large and complex, requiring specialized hardware to train and evaluate these models. To address such issues, we propose the PocketNet paradigm to reduce the size of deep learning models by throttling the growth of the number of channels in convolutional neural networks. We demonstrate that, for a range of segmentation and classification tasks, PocketNet architectures produce results comparable to that of conventional neural networks while reducing the number of parameters by multiple orders of magnitude, using up to 90% less GPU memory, and speeding up training times by up to 40%, thereby allowing such models to be trained and deployed in resource-constrained settings.

Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 23 of its 53 references that are in the map. First posted on arXiv 21 Apr 2021.

Builds on

23 of 53 references.

References · 23

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  3. 2016Densely Connected Convolutional Networkscited by 6,691
  4. 2017Squeeze-and-Excitation Networkscited by 4,763
  5. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,827
  6. 2016Xception: Deep Learning with Depthwise Separable Convolutionscited by 2,320
  7. 2016Identity Mappings in Deep Residual Networkscited by 1,960
  8. 2018UNet++: A Nested U-Net Architecture for Medical Image Segmentationcited by 1,518
  9. 20163D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotationcited by 1,471
  10. 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
  11. 2018A Convergence Theory for Deep Learning via Over-Parameterizationcited by 784
  12. 2017Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentationscited by 614
11 more
  1. 2018Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challengecited by 508
  2. 2019The Liver Tumor Segmentation Benchmark (LiTS)cited by 504
  3. 2020Overfitting in adversarially robust deep learningcited by 504
  4. 2018On the Optimization of Deep Networks: Implicit Acceleration by Overparameterizationcited by 280
  5. 2020COVID-Net: A Tailored Deep Convolutional Neural Network Design for Detection of COVID-19 Cases from Chest X-Ray Imagescited by 264
  6. 2016Efficient Multi-Scale 3D CNN with Fully Connected CRF for Accurate Brain Lesion Segmentationcited by 263
  7. 2020U-Net and its variants for medical image segmentation: theory and applicationscited by 169
  8. 2019Reducing the Hausdorff Distance in Medical Image Segmentation with Convolutional Neural Networkscited by 116
  9. 2019Multi-scale self-guided attention for medical image segmentationcited by 84
  10. 2019X-Net: Brain Stroke Lesion Segmentation Based on Depthwise Separable Convolution and Long-range Dependenciescited by 20
  11. 20183D Depthwise Convolution: Reducing Model Parameters in 3D Vision Taskscited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 4; 2024: 1; 2025: 3; 2026: 1.

9 papers.

  1. 2023A Generalized Surface Loss for Reducing the Hausdorff Distance in Medical Imaging Segmentationcited by 13
  2. 2024MIST: A Simple and Scalable End-To-End 3D Medical Imaging Segmentation Frameworkcited by 5
  3. 2023FMG-Net and W-Net: Multigrid Inspired Deep Learning Architectures For Medical Imaging Segmentationcited by 2
  4. 2023Solutions to Elliptic and Parabolic Problems via Finite Difference Based Unsupervised Small Linear Convolutional Neural Networkscited by 2
  5. 2023Joint inversion of Time-Lapse Surface Gravity and Seismic Data for Monitoring of 3D CO2 Plumes via Deep Learningcited by 1
  6. 2025Learning Discontinuous Galerkin Solutions to Elliptic Problems via Small Linear Convolutional Neural Networkscited by 1
  7. 2025AI and the Net-Zero Journey: Energy Demand, Emissions, and the Potential for Transitioncited by 0
  8. 2025Pre- and Post-Treatment Glioma Segmentation with the Medical Imaging Segmentation Toolkitcited by 0
  9. 2026Hybrid coupling with numerics-informed neural networks and the overlapping Schwarz alternating methodcited by 0