The official repository for "Revealing unforeseen diagnostic image features with deep learning by detecting cardiovascular diseases from apical four-chamber ultrasounds"

Overview

Revealing unforeseen diagnostic image features with deep learning by detecting cardiovascular diseases from apical four-chamber ultrasounds

image In this project, we aimed to develop a deep learning (DL) method to automatically detect impaired left ventricular (LV) function and aortic valve (AV) regurgitation from apical four-chamber (A4C) ultrasound cineloops. Two R(2+1)D convolutional neural networks (CNNs) were trained to detect the respective diseases. Subsequently, tSNE was used to visualize the embedding of the extracted feature vectors, and DeepLIFT was used to identify important image features associated with the diagnostic tasks.

The why

  • An automated echocardiography interpretation method requiring only limited views as input, say A4C, could make cardiovascular disease diagnosis more accessible.

    • Such system could become beneficial in geographic regions with limited access to expert cardiologists and sonographers.
    • It could also support general practitioners in the management of patients with suspected CVD, facilitating timely diagnosis and treatment of patients.
  • If the trained CNN can detect the diseases based on limited information, how?

    • Especially, AV regurgitation is typically diagnosed based on color Doppler images using one or more viewpoints. When given only the A4C view, would the model be able to detect regurgitation? If so, what image features does the model use to make the distinction? Since it’s on the A4C view, would the model identify some anatomical structure or movement associated with regurgitation, which are typically not being considered in conventional image interpretation? This is what we try to find out in the study.

Image features associated with the diagnostic tasks

DeepLIFT attributes a model’s classification output to certain input features (pixels), which allows us to understand which region or frame in an ultrasound is the key that makes the model classify it as a certain diagnosis. Below are some example analyses.

Representative normal cases

Case Averaged logit Input clip / Impaired LV function model's focus / AV regurgitation model's focus
Normal1 0.9999 image
Normal2 0.9999 image
Normal3 0.9999 image
Normal4 0.9999 image
Normal5 0.9999 image
Normal6 0.9999 image
Normal7 0.9998 image
Normal8 0.9998 image
Normal9 0.9998 image
Normal10 0.9997 image

DeepLIFT analyses reveal that the LV myocardium and mitral valve were important for detecting impaired LV function, while the tip of the mitral valve anterior leaflet, during opening, was considered important for detecting AV regurgitation. Apart from the above examples, all confident cases are provided, which the predicted probability of being the normal class by the two models are both higher than 0.98. See the full list here.

Representative disease cases

  • Mildly impaired LV
Case Logit Input clip / Impaired LV function model's focus
MildILV1 0.9989 image
MildILV2 0.9988 image
  • Severely impaired LV
Case Logit Input clip / Impaired LV function model's focus
SevereILV1 1.0000 image
SevereILV2 1.0000 image
  • Mild AV regurgitation
Case Logit Input clip / AV regurgitation model's focus
MildAVR1 0.7240 image
MildAVR2 0.6893 image
  • Substantial AV regurgitation
Case Logit Input clip / AV regurgitation model's focus
SubstantialAVR1 0.9919 image
SubstantialAVR2 0.9645 image

When analyzing disease cases, the highlighted regions in different queries are quite different. We speculate that this might be due to a higher heterogeneity in the appearance of the disease cases. Apart from the above examples, more confident disease cases are provided. See the full list here.

Run the code on your own dataset

The dataloader in util can be modified to fit your own dataset. To run the full workflow, namely training, validation, testing, and the subsequent analyses, simply run the following commands:

git clone https://github.com/LishinC/Disease-Detection-and-Diagnostic-Image-Feature.git
cd Disease-Detection-and-Diagnostic-Image-Feature/util
pip install -e .
cd ../projectDDDIF
python main.py

Loading the trained model weights

The model weights are made available for external validation, or as pretraining for other echocardiography-related tasks. To load the weights, navigate to the projectDDDIF folder, and run the following python code:

import torch
import torch.nn as nn
import torchvision

#Load impaired LV model
model_path = 'model/impairedLV/train/model_val_min.pth'
# #Load AV regurgitation model
# model_path = 'model/regurg/train/model_val_min.pth'

model = torchvision.models.video.__dict__["r2plus1d_18"](pretrained=False)
model.stem[0] = nn.Conv3d(1, 45, kernel_size=(1, 7, 7), stride=(1, 2, 2), padding=(0, 3, 3), bias=False)
model.fc = nn.Linear(model.fc.in_features, 3)
model.load_state_dict(torch.load(model_path))

Questions and feedback

For techinical problems or comments about the project, feel free to contact [email protected].

This is an official implementation for "Video Swin Transformers".

Video Swin Transformer By Ze Liu*, Jia Ning*, Yue Cao, Yixuan Wei, Zheng Zhang, Stephen Lin and Han Hu. This repo is the official implementation of "V

Swin Transformer 981 Jan 03, 2023
This repository contains code released by Google Research.

This repository contains code released by Google Research.

Google Research 26.6k Dec 31, 2022
Music source separation is a task to separate audio recordings into individual sources

Music Source Separation Music source separation is a task to separate audio recordings into individual sources. This repository is an PyTorch implmeme

Bytedance Inc. 958 Jan 03, 2023
Graph-based community clustering approach to extract protein domains from a predicted aligned error matrix

Using a predicted aligned error matrix corresponding to an AlphaFold2 model , returns a series of lists of residue indices, where each list corresponds to a set of residues clustering together into a

Tristan Croll 24 Nov 23, 2022
Repo for CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learning

CReST in Tensorflow 2 Code for the paper: "CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learning" by Chen Wei, Ki

Google Research 75 Nov 01, 2022
Minimal implementation of Denoised Smoothing: A Provable Defense for Pretrained Classifiers in TensorFlow.

Denoised-Smoothing-TF Minimal implementation of Denoised Smoothing: A Provable Defense for Pretrained Classifiers in TensorFlow. Denoised Smoothing is

Sayak Paul 19 Dec 11, 2022
QKeras: a quantization deep learning library for Tensorflow Keras

QKeras github.com/google/qkeras QKeras 0.8 highlights: Automatic quantization using QKeras; Stochastic behavior (including stochastic rouding) is disa

Google 437 Jan 03, 2023
TargetAllDomainObjects - A python wrapper to run a command on against all users/computers/DCs of a Windows Domain

TargetAllDomainObjects A python wrapper to run a command on against all users/co

Podalirius 19 Dec 13, 2022
MolRep: A Deep Representation Learning Library for Molecular Property Prediction

MolRep: A Deep Representation Learning Library for Molecular Property Prediction Summary MolRep is a Python package for fairly measuring algorithmic p

AI-Health @NSCC-gz 83 Dec 24, 2022
Code repository for "Free View Synthesis", ECCV 2020.

Free View Synthesis Code repository for "Free View Synthesis", ECCV 2020. Setup Install the following Python packages in your Python environment - num

Intelligent Systems Lab Org 253 Dec 07, 2022
ColBERT: Contextualized Late Interaction over BERT (SIGIR'20)

Update: if you're looking for ColBERTv2 code, you can find it alongside a new simpler API, in the branch new_api. ColBERT ColBERT is a fast and accura

Stanford Future Data Systems 637 Jan 08, 2023
A curated list of Generative Deep Art projects, tools, artworks, and models

Generative Deep Art A curated list of Generative Deep Art projects, tools, artworks, and models Inbox Get started with making AI art in 2022 – deeplea

Filipe Calegario 251 Jan 03, 2023
StableSims is an open-source project aimed at simulating MakerDAO's Dai stablecoin system

StableSims is an open-source project aimed at simulating MakerDAO's Dai stablecoin system, initially used for researching optimal incentive parameters for Liquidations 2.0.

Blockchain at Berkeley 52 Nov 21, 2022
BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search

BossNAS This repository contains PyTorch evaluation code, retraining code and pretrained models of our paper: BossNAS: Exploring Hybrid CNN-transforme

Changlin Li 127 Dec 26, 2022
Notebooks for my "Deep Learning with TensorFlow 2 and Keras" course

Deep Learning with TensorFlow 2 and Keras – Notebooks This project accompanies my Deep Learning with TensorFlow 2 and Keras trainings. It contains the

Aurélien Geron 1.9k Dec 15, 2022
Application of the L2HMC algorithm to simulations in lattice QCD.

l2hmc-qcd 📊 Slides Recent talk on Training Topological Samplers for Lattice Gauge Theory from the Machine Learning for High Energy Physics, on and of

Sam Foreman 37 Dec 14, 2022
Pytorch implementation of the paper "Class-Balanced Loss Based on Effective Number of Samples"

Class-balanced-loss-pytorch Pytorch implementation of the paper Class-Balanced Loss Based on Effective Number of Samples presented at CVPR'19. Yin Cui

Vandit Jain 697 Dec 29, 2022
code for EMNLP 2019 paper Text Summarization with Pretrained Encoders

PreSumm This code is for EMNLP 2019 paper Text Summarization with Pretrained Encoders Updates Jan 22 2020: Now you can Summarize Raw Text Input!. Swit

Yang Liu 1.2k Dec 28, 2022
This program writes christmas wish programmatically. It is using turtle as a pen pointer draw christmas trees and stars.

Introduction This is a simple program is written in python and turtle library. The objective of this program is to wish merry Christmas programmatical

Gunarakulan Gunaretnam 1 Dec 25, 2021