Mass detection in mammograms using pre-trained deep learning models

For the purpose of accurate and efficient mass detection in full-field digital mammograms, weproposeamethodfor automated mass detection that consists oftwostages: suspicious region localization and false-positive (FP) reduction, by classifying these regions into mass and non-mass regions (normal tissues). In the first stage, the density of the wavelet coefficients based on Quincunx Lifting Scheme (DWC-QLS) is used to find suspicious regions (regions of interest, ROIs) in full mammograms. In the second stage, a patch-based CNN classifier is developed as an FP reduction to classify the suspicious regions. The main aim of this stage is to reduce the false-positive suspicious regions while keeping the true-positive suspicious regions. To further improve the performance of the FP reduction, the effectiveness of different transfer learning strategies is further explored and the best fine-tuning strategy in training InceptionV3 model is determined experimentally. The experimental results show that the proposed method can achieve an overall performance of 0.98 TPR @1.43 FPI on the INbreast database. In addition, the suggested segmentation method detects the mass location with 100% sensitivity and average of 5.4 false positives per image. Based on the obtained results,the introduced method was able to successfully detectand classify suspicious regions in digital mammograms and provide better TPR and FPI results in comparison with other state-of-the-art method.

In deep convolutional neural networks (DCNNs), model size and computation complexity are two important factors governing throughput and energy efficiency when deployed to hardware for inference. Recent works on compact DCNNs as well as pruning methods are effective, yet with drawbacks. For instance, more than half the size of all MobileNet models lies in their last two layers, mainly because compact separable convolution (CONV) layers are not applicable to their last fully connected (FC) layers. Also, in pruning methods, the compression is gained at the expense of irregularity in the DCNN architecture, which necessitates additional indexing memory to address nonzero weights, thereby increasing memory footprint, decompression delays, and energy consumption. In this article, we propose cyclic sparsely connected (CSC) architectures, with memory/computation complexity of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathcal {O}(N\log {}N)$ </tex-math></inline-formula> , where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$N$ </tex-math></inline-formula> is the number of nodes/channels given a DCNN layer that, contrary to compact depthwise separable layers, can be used as an overlay for both FC and CONV layers of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathcal {O}(N^{2})$ </tex-math></inline-formula> . Also, contrary to pruning methods, CSC architectures are structurally sparse and require no indexing due to their cyclic nature. We show that both standard convolution and depthwise convolution layers are special cases of the CSC layers, whose mathematical function, along with FC layers, can be unified into one single formulation and whose hardware implementation can be carried out under one arithmetic logic component. We examine the efficacy of the CSC architectures for compression of LeNet, AlexNet, and MobileNet DCNNs with precision ranging from 2 to 32 bits. More specifically, we surge upon the compact 8-bit quantized 0.5 MobileNet V1 and show that by compressing its last two layers with CSC architectures, the model is compressed by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim 1.5\times $ </tex-math></inline-formula> with a size of only 873 kB and little accuracy loss. Finally, we design a configurable hardware that implements all types of DCNN layers including FC, CONV, depthwise, CSC-FC, and CSC-CONV indistinguishably within a unified pipeline. We implement the hardware on a tiny Xilinx field-programmable gate array (FPGA) for total on-chip processing of the compressed MobileNet that, compared to the related work, has the highest Inference/J while utilizing the smallest FPGA.

The deep convolutional neural networks (CNN) have a vast amount of parameters, especially in the fully connected (FC) layers, which have become a bottleneck for real-time applications where processing latency is high due to computational cost. In this paper, we propose to optimize the FC layers in CNN via making it much slimmer. We make analysis of the statistical distribution of the weights in FC layer, and observe each column follows Gaussian distribution. Regression model analysis of the weights of FC layer based on Akaike information criteria and Bayesian information criterion demonstrates that they have Granger causality, which means the columns are correlated and they follow colored Gaussian distribution. Based on this distribution, we derive a CNN design and optimization theorem for FC layers from information theory point of view. The theorem provides two design criteria, rank and singular values. Further, we show that FC layer with weights of colored Gaussian is more efficient than that of white Gaussian. The optimization criteria is singular-values-based, so we apply singular value decomposition to find the maximal singular values and QR to identify the corresponding columns in FC layer. We evaluate our optimization approach to AlexNet and apply the slimmer CNN to ImageNet classification. Simulation results show our approach performs much better than random dropout. Specifically, with only around 28% of weights, the AlexNet could perform as well as the original AlexNet in terms of top one error and top five error.

Identifying sleep problem severity from overnight polysomnography (PSG) recordings plays an important role in diagnosing and treating sleep disorders such as the Obstructive Sleep Apnea (OSA). This analysis traditionally is done by specialists manually through visual inspections, which can be tedious, time-consuming, and is prone to subjective errors. One of the solutions is to use Convolutional Neural Networks (CNN) where the convolutional and pooling layers behave as feature extractors and some fully-connected (FCN) layers are used for making final predictions for the OSA severity. In this paper, a CNN architecture with 1D convolutional and FCN layers for classification is presented. The PSG data for this project are from the Cleveland Children’s Sleep and Health Study database and classification results confirm the effectiveness of the proposed CNN method. The proposed 1D CNN model achieves excellent classification results without manually preprocesssing PSG signals such as feature extraction and feature reduction.

Breast cancer is one of the most leading causes of cancer deaths among women. Early detection of cancer increases the survival rate of the affected women. Machine learning approaches that are used for classification of breast cancer usually takes a lot of processing time during the training process. This paper attempts to propose a Machine Learning approach for breast cancer detection in mammograms, which does not depend on the number of training samples. The paper aims to develop a core vector machine-based diagnosis system for breast cancer detection using the date from MIAS. The main motivation behind using this system is to reduce the computational and memory requirement for large training data and to improve the classification accuracy. The proposed method has four stages: 1) Pre-processing is done to extract the breast region using global thresholding and enhancement using histogram equalization; 2) identification of potential mass using Otsu thresholding; 3) feature extraction using Laws Texture energy measures; and 4) mass detection is done using Core vector machine (CVM) classifier. Comparative analysis was done with different existing algorithms: Artificial Neural Network (ANN), Support Vector Machine (SVM), and Fuzzy Support Vector Machines (FSVM). The results illustrate that the proposed Core Vector Machine (CVM) classifier produced a promising result in terms of sensitivity (96.9%), misclassification rate (0.0443) and accuracy (95.89%). The time taken for training process is 0.0443, which is less when compared with other machine learning algorithms. Performance analysis shows that CVM classifier is superior to other classifiers like ANN, SVM and FSVM. The computational time of the CVM classifier during the training process was also analysed and found to be better than other discussed algorithms. The results achieved show that CVM classifier is the best algorithm for breast mass detection in mammograms.

Convolutional neural network with transfer learning approach for detection of unfavorable driving state using phase coherence image

Detection of suspicious masses in mammograms play a vital role in early diagnosis of breast cancer, to reduce the death rate among women. The presence of masses and calcification’s in mammograms is distinguishing signs for breast cancer diagnosis. But in some cases, due to contrast variation, fuzzy boundaries and presence of noise in mammograms, segmentation and detection of masses is challenging assignment. This paper presents a new segmentation approach to detect masses in mammographic images. The proposed approach consist of artifacts elimination and pectoral region extraction, suspicious mass enhancement using dual morphological operation technique and finally, extraction of Regions of Interest (ROIs) from background using scaled Reyni entropy. The proposed system has been tested on two data-sets i.e. mini- Mammographic Image Analysis Society (mini-MIAS) and Digital Database for Screening Mammography (DDSM), over 50 and 90 mammograms respectively. Performance achieved with proposed system in terms of True Positive Fraction (TPF) yields 93.2% and 93.9% respectively, at the rate of 1.48 and 0.74 average False Positive per Image (FP/I), tested on both Cranio-Caudal (CC) and Medio-Lateral Oblique (MLO) views. The obtained experimental results demonstrates that proposed method gives improved results for mass detection and can be useful for radiologists in diagnosis of breast cancer at early stage.

An architecture for automatic lung tissue classification method based on the Deep Learning techniques is designed in this paper. Recent works on Deep Learning techniques achieved impressive results in the field of medical image classification. So, we designed a Convolution Neural Network (CNN) for the classification of five categories of Interstitial Lung Diseases (ILD) patterns in High-Resolution Computed Tomography (HRCT) images. The CNN consists of 3 Convolution layers, Leaky ReLU activation followed by Maximum pooling layer and dense layer. The last Fully Connected (FC) layer has 5 outputs equivalent to the classes considered such as Normal, Ground Glass (GG), Emphysema, Micro Nodules, and Fibrosis. The proposed CNN is trained and evaluated on the publicly available ILD database provided by the University Hospitals of Geneva (HUG). Experimental results are compared with the state-of-art, which shows an outstanding performance of the proposed CNN model giving 94.67% accuracy and 94.65% Favg .

The enormous volume of data, the reduction of human error, and the high classification throughput have made the automation of blood cell cancer image classification and recognition more and more popular. This work uses a number of deep learning (DL) models from Convolutional Neural Networks (CNNs) to test the classification accuracy. Alexnet, Resnet50, Visual Geometry Group 16 (VGG16), Visual Geometry Group 19 (VGG19), and GoogLeNet are the models. Various convolutional layers, including the Fully Connected (FC), max pooling, and soft-max layers, make up the architecture of these models. Since Vgg19 model showed the highest accuracy of classification with epoch size= 30, Minimum Batch Size (MBS)= 64, learning rate= 0.001 and Sgdm optimizer, it has been modified for less complexity and higher performance than its based model. The confusion matrix is used to assess and display the performance of the base models and the modified Vgg19 model. Four types of blood cell cancered images were used in this work, these are: Eosinophil, Neutrophil, Eosinophil and Monocyte. Modified Vgg19 achieved an accuracy of 0.983 and complexity in term of FLOPs is reduced to 0.7% of the original complexity for Vgg19 based model.

Breast cancer is the second most common type of cancer in the world. Several computer-aided detection and diagnosis systems have been used to assist health experts and to indicate suspect areas that would be difficult to perceive by the human eye; this approach has aided in the detection and diagnosis of cancer. The present work proposes a method for the automatic detection of masses in digital mammograms by using quality threshold (QT), a correlogram function, and the support vector machine (SVM). This methodology comprises the following steps: The first step is to perform preprocessing with a low-pass filter, which increases the scale of the contrast, and the next step is to use an enhancement to the wavelet transform with a linear function. After the preprocessing is segmentation using QT; then, we perform post-processing, which involves the selection of the best mass candidates. This step is performed by analyzing the shape descriptors through the SVM. For the stage that involves the extraction of texture features, we used Haralick descriptors and a correlogram function. In the classification stage, the SVM was again used for training, validation, and final test. The results were as follows: sensitivity 92.31 %, specificity 82.2 %, accuracy 83.53 %, mean rate of false positives per image 1.12, and area under the receiver operating characteristic (ROC) curve 0.8033. Breast cancer is notable for presenting the highest mortality rate in addition to one of the smallest survival rates after diagnosis. An early diagnosis means a considerable increase in the survival chance of the patients. The methodology proposed herein contributes to the early diagnosis and survival rate and, thus, proves to be a useful tool for specialists who attempt to anticipate the detection of masses.

The application of deep learning technology has increased rapidly in recent years. Technologies in deep learning increasingly emulate natural human abilities, such as knowledge learning, problem-solving, and decision-making. In general, deep learning can carry out self-training without repetitive programming by humans. Convolutional neural networks (CNNs) are deep learning algorithms commonly used in wide applications. CNN is often used for image classification, segmentation, object detection, video processing, natural language processing, and speech recognition. CNN has four layers: convolution layer, pooling layer, fully connected layer, and non-linear layer. The convolutional layer uses kernel filters to calculate the convolution of the input image by extracting the fundamental features. The pooling layer combines two successive convolutional layers. The third layer is the fully connected layer, commonly called the convolutional output layer. The activation function defines the output of a neural network, such as 'yes' or 'no'. The most common and popular CNN activation functions are Sigmoid, Tanh, ReLU, Leaky ReLU, Noisy ReLU, and Parametric Linear Units. The organization and function of the visual cortex greatly influence CNN architecture because it is designed to resemble the neuronal connections in the human brain. Some of the popular CNN architectures are LeNet, AlexNet and VGGNet.

Skin Cancer is an extensive and possibly dangerous disorder that requires early detection for effective treatment. Add specific global statistics on skin cancer prevalence and mortality to emphasize the importance of early detection. Example: “Skin cancer accounts for 1 in 5 diagnosed cancers globally, with melanoma causing over 60,000 deaths annually. Manual skin cancer screening is both time-intensive and expensive. Deep learning (DL) techniques have shown exceptional performance in various applications and have been applied to systematize skin cancer diagnosis. However, training DL models for skin cancer diagnosis is challenging due to limited available data and the risk of overfitting. Traditionally approaches have High computational costs, a lack of interpretability, deal with numerous hyperparameters and spatial variation have always been problems with machine learning (ML) and DL. An innovative method called adaptive learning has been developed to overcome these problems. In this research, we advise an intelligent computer-aided system for automatic skin cancer diagnosis using a two-stage transfer learning approach and Pre-trained Convolutional Neural Networks (CNNs). CNNs are well-suited for learning hierarchical features from images. Annotated skin cancer photographs are utilized to detect ROIs and reset the initial layer of the pre-trained CNN. The lower-level layers learn about the characteristics and patterns of lesions and unaffected areas by fine-tuning the model. To capture high-level, global features specific to skin cancer, we replace the fully connected (FC) layers, responsible for encoding such features, with a new FC layer based on principal component analysis (PCA). This unsupervised technique enables the mining of discriminative features from the skin cancer images, effectively mitigating overfitting concerns and letting the model adjust structural features of skin cancer images, facilitating effective detection of skin cancer features. The system shows great potential in facilitating the initial screening of skin cancer patients, empowering healthcare professionals to make timely decisions regarding patient referrals to dermatologists or specialists for further diagnosis and appropriate treatment. Our advanced adaptive fine-tuned CNN approach for automatic skin cancer diagnosis offers a valuable tool for efficient and accurate early detection. By leveraging DL and transfer learning techniques, the system has the possible to transform skin cancer diagnosis and improve patient outcomes.

1 - Congruence of deep learning in biomedical engineering: future prospects and challenges

In today's scenario Breast Cancer is one of the most prominent causes for demise of a female universally, which increases the demand of technologies on war footings providing premature and accurate detection in order to be efficiently treated. Hence to satisfy the same the mammography has emerged as one of the most popular detecting methods, but because of the high inter-observer variability and understated irregularities, it found difficult to construe mammograms. In this quest, we hereby are proposing a framework for automated breast cancer detection in mammogram images that depends on 2D Convolutional Neural Networks (CNN). The suggested model makes use of CNNs' hierarchical feature of extraction capabilities to accurately diagnose malignant lesions. We achieved outmost results with respect to reliability and accuracy by training & evaluating model with publicly available data. We experimented these methods on Mammography Breast Cancer Detection (MBCD) data. The outcomes of the studies show that the 2D CNN architecture minimizes false positives and, maximizes diagnostic reliability by successfully capturing intricate patterns in mammograms. By providing a trustworthy alternative opinion, the above approach may assist clinicians in increasing early identification rates and improving the treatment of patients. The investigation focuses on how deep learning is transforming medical imaging and how easy it is to use these kinds of systems in healthcare settings.

Recent years, convolutional neural network (CNN) is a research hot spot in machine learning and has some application value in computer aided diagnosis. Firstly, this paper briefly introduces the basic principle of CNN. Secondly, it summarizes the improvement on network structure from two dimensions of model and structure optimization. In model structure, it summarizes eleven classical models about CNN in the past 60 years, and introduces its development process according to timeline. In structure optimization, the research progress is summarized from five aspects (input layer, convolution layer, down-sampling layer, full-connected layer and the whole network) of CNN. Thirdly, the learning algorithm is summarized from the optimization algorithm and fusion algorithm. In optimization algorithm, it combs the progress of the algorithm according to optimization purpose. In algorithm fusion, the improvement is summarized from five angles: input layer, convolution layer, down-sampling layer, full-connected layer and output layer. Finally, CNN is mapped into the medical image domain, and it is combined with computer aided diagnosis to explore its application in medical images. It is a good summary for CNN and has positive significance for the development of CNN.