-
Notifications
You must be signed in to change notification settings - Fork 1
Home
Ethan Weinberger edited this page May 31, 2018
·
24 revisions
- Figure out how to export qpdata annotations into a format that python can read
- Official Tensorflow transfer learning tutorial: https://www.tensorflow.org/tutorials/image_retraining
- Pretrained model files: https://github.com/tensorflow/models/tree/master/research/slim
- Focus on Resnet/Inception variants
- Functions for data augmentation: https://medium.com/ymedialabs-innovation/data-augmentation-techniques-in-cnn-using-tensorflow-371ae43d5be9
-
H&E-stained Whole Slide Image Deep Learning Predicts SPOP
Mutation State in Prostate Cancer (https://www.biorxiv.org/content/biorxiv/early/2017/05/12/064279.full.pdf)
- Goal: Use deep learning to predict SPOP gene mutations in prostate cancer cells
- Used an ensemble of ResNet-50 models modified with an additional dropout + fully connected layer.
- Networks were all pretrained on ImageNet data.
- Each network in the ensemble was trained on a small draw of slide data (had 157 patient slides in total).
- Weak networks (weak since they were trained on little data) were combined to form a strong learner ensemble
- Caveat: Paper doesn't provide indication of how well/poorly trained pathologists perform on this data
-
A deep-learning classifier identifies patients
with clinical heart failure using whole-slide
images of H&E tissue (http://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0192726&type=printable)
- Goal: Identify patients with clinical heart failure using H&E tissue
- Dataset: 209 patients
- 94 with heart failure
- 115 w/o failure
- Compared CNN to traditional feature-engineering model
- Pathologists classify correctly with 75% accuracy - CNNs beat this with 20% increase on sensitivity and specificity
- Architecture was modified from that of Janowczyk and Madabhushi (these guys do lots of DL pathology work)
-
Classification of breast cancer histology images
using transfer learning (https://arxiv.org/pdf/1802.09424.pdf)
- Investigated the transfer learning performance of InceptionV3 and ResNet-50 with H&E images
- Specifically studied a breast cancer cell classification task
- Found that ResNet's performance was better than that of Inception
- Investigated the transfer learning performance of InceptionV3 and ResNet-50 with H&E images
-
Transfer Learning for Cell Nuclei Classification
in Histopathology Images (https://pdfs.semanticscholar.org/2069/fd9089ad9eb8d0272715e24eed94fd9b1813.pdf)
- Older work on transfer learning for histopathology applications
- Paper is from 2016 and only uses older architectures (GoogLeNet, Alexnet, etc) so probably not too relevant