Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MIST: A Simple and Scalable End-To-End 3D Medical Imaging Segmentation Framework

Adrian Celaya, Evan Lim, Rachel Glenn, Brayden Mi, Alex Balsells, Dawid Schellingerhout, Tucker Netherton, Caroline Chung, Beatrice Riviere, David Fuentes

TL;DR

The Medical Imaging Segmentation Toolkit (MIST) is introduced, a simple, modular, and end-to-end medical imaging segmentation framework designed to facilitate consistent training, testing, and evaluation of deep learning-based medical imaging segmentation methods.

Abstract

Medical imaging segmentation is a highly active area of research, with deep learning-based methods achieving state-of-the-art results in several benchmarks. However, the lack of standardized tools for training, testing, and evaluating new methods makes the comparison of methods difficult. To address this, we introduce the Medical Imaging Segmentation Toolkit (MIST), a simple, modular, and end-to-end medical imaging segmentation framework designed to facilitate consistent training, testing, and evaluation of deep learning-based medical imaging segmentation methods. MIST standardizes data analysis, preprocessing, and evaluation pipelines, accommodating multiple architectures and loss functions. This standardization ensures reproducible and fair comparisons across different methods. We detail MIST's data format requirements, pipelines, and auxiliary features and demonstrate its efficacy using the BraTS Adult Glioma Post-Treatment Challenge dataset. Our results highlight MIST's ability to produce accurate segmentation masks and its scalability across multiple GPUs, showcasing its potential as a powerful tool for future medical imaging research and development.

MIST: A Simple and Scalable End-To-End 3D Medical Imaging Segmentation Framework

TL;DR

The Medical Imaging Segmentation Toolkit (MIST) is introduced, a simple, modular, and end-to-end medical imaging segmentation framework designed to facilitate consistent training, testing, and evaluation of deep learning-based medical imaging segmentation methods.

Abstract

Medical imaging segmentation is a highly active area of research, with deep learning-based methods achieving state-of-the-art results in several benchmarks. However, the lack of standardized tools for training, testing, and evaluating new methods makes the comparison of methods difficult. To address this, we introduce the Medical Imaging Segmentation Toolkit (MIST), a simple, modular, and end-to-end medical imaging segmentation framework designed to facilitate consistent training, testing, and evaluation of deep learning-based medical imaging segmentation methods. MIST standardizes data analysis, preprocessing, and evaluation pipelines, accommodating multiple architectures and loss functions. This standardization ensures reproducible and fair comparisons across different methods. We detail MIST's data format requirements, pipelines, and auxiliary features and demonstrate its efficacy using the BraTS Adult Glioma Post-Treatment Challenge dataset. Our results highlight MIST's ability to produce accurate segmentation masks and its scalability across multiple GPUs, showcasing its potential as a powerful tool for future medical imaging research and development.
Paper Structure (29 sections, 3 figures, 3 tables)

This paper contains 29 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Data Format
  4. Pipelines
  5. Data Analysis
  6. Cropping to Foreground
  7. Target Spacing
  8. Patch Size Selection
  9. Normalization Parameters
  10. Other Checks
  11. Preprocessing
  12. Reorient & Resample
  13. Window & Normalize
  14. Distance Transform Maps
  15. Training
  16. ...and 14 more sections

Figures (3)

  • Figure 1: Workflow for the MIST preprocessing pipeline. This preprocessing pipeline takes raw NIfTI files and outputs cropped (optional), bias corrected (optional and MR only), reoriented, resampled, windowed, and normalized NumPy files. This pipeline can also compute the DTMs for each label in the ground truth mask and output them as NumPy files.
  • Figure 2: Time per epoch for different batch sizes and numbers of A100 GPUs. Here, we see that MIST achieves better scaling with larger batch sizes for A100 GPUs.
  • Figure 3: Time per epoch for different batch sizes and numbers of H100 GPUs. Here, we see that MIST achieves near optimal scaling for all batch sizes.