Skip to main content

Deep Learning at Chest Radiography: Automated Classification of Pulmonary Tuberculosis by Using Convolutional Neural Networks

In this paper, the authors explore the use of Deep Convolutional Neural Networls (DCNN) in classifying Tuberculosis (TB) in chest radiographs. One of the advantages of deep learning is its ability to excel with high dimensional datasets, such as images, which can be represented at multiple levels. 

Dataset

Four deidentified HIPAA-compliant datasets were used in this study that were exempted from review by the institutional review board, which consisted of 1007 posteroanterior chest radiographs.

DCNN Models and Training

AlexNet and GoogLeNet models, including pre-trained (on ImageNet from Caffe Model Zoo) and untrained models were used in the study. It was found that the AUCs of the pretrained networks were greater. The following solver parameters were used for training: 120 epochs; base learning rate for untrained models and for pretrained models, 0.01 and 0.001, respectively with stochastic gradient descent.  Both of the DCNNs in this studied used dropout or model regularization strategies to help overcome overfitting.

Data Augmentation

The following data augmentation techniques further increased the performance

  1. Random cropping of 227x227
  2. Mean subtraction and mirror images
  3. Rotation of 90, 180 and 270.
  4. Contrast Limited Adaptive Histogram Equalization processing

Ensembling


The ensembling technique was used to increase the AUC even further. Ensembles were performed by taking different weighted averages of the probability scores generated by the classifiers 
The best performing ensemble model had an AUC of 0.99. Refer below the table borrowed from the paper for complete results

 The sensitivity of pre-trained AlexNet was 92.0% and the specificity was 94.7%. The sensitivity of pre-trained GoogLeNet was 92.0% and the specificity was 98.7%. The sensitivity of the ensemble was 97.3% and the specificity was 94.7%. 

Radiologist-augmented approach

This is were the paper takes turn to beyond the realms of deep learning, were they use a certain human to classify the images were the models fail.

 For cases where the AlexNet and GoogLeNet classifiers had disagreement, an independent board-certified cardiothoracic radiologist (B.S., with 18 years of experience) blindly interpreted the images as either having manifestations of TB or as normal. This resulted in a sensitivity of 97.3% and a specificity of 100%.

Comments

Post a Comment

Popular Posts

Network In Network

In this paper , the authors introduce a new network structure for the traditional CNN to better extract and interpret latent features. It is named "Network In Network (NIN)". NIN vs tradional CNN In a traditional CNN, convolutional layers and spatial pooling layers are stacked followed by fully connected layers and an output layer. The convolution layers generate feature maps by linear convolutional filters followed by non-linear activation functions. The NIN structure addresses the following 2 limitations of a traditional CNN. Kernels/filters used for each CNN layer works well when the features to be extracted are linearly separable. Fully connected layers at the end of the CNN leads to over-fitting the training data.  Convolution with linear filter vs Neural network The convolution layers involve a kernel that slides over the previous field (input or layers) and extracts features. The kernel is usually a matrix with which convolution is done. This is a linear operation. Mea...
Post a Comment
Read more

Joint Pose and Shape Estimation of Vehicles from LiDAR Data

In this paper , the authors address the problem of estimating the pose and shape of vehicles from LiDAR Data. This is a common problem to be solved in autonomous vehicle applications. Autonomous vehicles are equipped with many sensors to perceive the world around them. LiDAR being one of them is what the authors focus on in this paper. A key requirement of the perception system is to identify other vehicles in the road and make decisions based on their pose and shape. The authors put forth a pipeline that jointly determines pose and shape from LiDAR data.  More about Pose and Shape Estimation LiDAR sensors capture the world around them in point clouds. Often, the first step in LiDAR processing is to perform some sort of clustering or segmentation, to isolate parts of the point cloud which belong to individual objects.  The next step is to infer the pose and shape of the object. This is mostly done by a modal perception . Meaning the whole object is perceived based on partial s...
Post a Comment
Read more

BLIP: Bootstrapping Language-Image Pretraining for Unified Vision-Language Understanding

BLIP is a new vision-language model proposed by Microsoft Research Asia in 2022. It introduces a bootstrapping method to learn from noisy image-text pairs scraped from the web. The BLIP Framework BLIP consists of three key components: MED  - A multimodal encoder-decoder model that can encode images, text, and generate image-grounded text. Captioner  - Fine-tuned on COCO to generate captions for web images. Filter  - Fine-tuned on COCO to filter noisy image-text pairs. The pretraining process follows these steps: Collect noisy image-text pairs from the web. Pretrain MED on this data. Finetune captioner and filter on the COCO dataset. Use captioner to generate new captions for web images. Filter noisy pairs using the filter model. Repeat the process by pretraining on a cleaned dataset. This bootstrapping allows BLIP to learn from web-scale noisy data in a self-supervised manner. Innovations in BLIP Some interesting aspects of BLIP: Combines encoder-decoder capability in one...
Post a Comment
Read more