Lung Pattern Classification for Interstitial Lung Diseases Using a Deep Convolutional Neural Network

Automated tissue characterization is one of the most crucial components of a computer aided diagnosis (CAD) system for interstitial lung diseases (ILDs). Although much research has been conducted in this field, the problem remains challenging. Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. In this paper, we propose and evaluate a convolutional neural network (CNN), designed for the classification of ILD patterns. The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average pooling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation. To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals. To the best of our knowledge, this is the first deep CNN designed for the specific problem. A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans and integrating the proposed method into a CAD system that aims to provide differential diagnosis for ILDs as a supportive tool for radiologists.

Evaluating retraining rules for semi-supervised learning in neural network based cursive word recognition

Training a system to recognize handwritten words is a task that requires a large amount of data with their correct transcription. However, the creation of such a training set, including the generation of the ground truth, is tedious and costly. One way of reducing the high cost of labeled training data acquisition is to exploit unlabeled data, which can be gathered easily. Making use of both labeled and unlabeled data is known as semi-supervised learning. One of the most general versions of semi-supervised learning is self-training, where a recognizer iteratively retrains itself on its own output on new, unlabeled data. In this paper we propose to apply semi-supervised learning, and in particular self-training, to the problem of cursive, handwritten word recognition. The special focus of the paper is on retraining rules that define what data are actually being used in the retraining phase. In a series of experiments it is shown that the performance of a neural network based recognizer can be significantly improved through the use of unlabeled data and self-training if appropriate retraining rules are applied.

Comparison of dual-energy subtraction and electronic bone suppression combined with computer-aided detection on chest radiographs: effect on human observers' performance in nodule detection

The objective of our study was to compare the effect of dual-energy subtraction and bone suppression software alone and in combination with computer-aided detection (CAD) on the performance of human observers in lung nodule detection.

The performance of computer-aided detection when analyzing prior mammograms of newly detected breast cancers with special focus on the time interval from initial imaging to detection

PURPOSE: The clinical role of CAD systems to detect breast cancer, which have not been on cancer containing mammograms not detected by the radiologist was proven retrospectively. METHODS: All patients from 1992 to 2005 with a histologically verified malignant breast lesion and a mammogram at our department, were analyzed in retrospect focussing on the time of detection of the malignant lesion. All prior mammograms were analyzed by CAD (CADx, USA). The resulting CAD printout was matched with the cancer containing images yielding to the radiological diagnosis of breast cancer. CAD performance, sensitivity as well as the association of CAD and radiological features were analyzed. RESULTS: 278 mammograms fulfilled the inclusion criteria. 111 cases showed a retrospectively visible lesion (71 masses, 23 single microcalcification clusters, 16 masses with microcalcifications, in one case two microcalcification clusters). 54/87 masses and 34/41 microcalcifications were detected by CAD. Detection rates varied from 9/20 (ACR 1) to 5/7 (ACR 4) (45% vs. 71%). The detection of microcalcifications was not influenced by breast tissue density. CONCLUSION: CAD might be useful in an earlier detection of subtle breast cancer cases, which might remain otherwise undetected.

Food Recognition for Dietary Assessment Using Deep Convolutional Neural Networks

Diet management is a key factor for the prevention and treatment of diet-related chronic diseases. Computer vision systems aim to provide automated food intake assessment using meal images. We propose a method for the recognition of already segmented food items in meal images. The method uses a 6-layer deep convolutional neural network to classify food image patches. For each food item, overlapping patches are extracted and classified and the class with the majority of votes is assigned to it. Experiments on a manually annotated dataset with 573 food items justified the choice of the involved components and proved the effectiveness of the proposed system yielding an overall accuracy of 84.9%.

Lung cancer screening with CT: evaluation of radiologists and different computer assisted detection software (CAD) as first and second readers for lung nodule detection at different dose levels

OBJECTIVES To find the best pairing of first and second reader at highest sensitivity for detecting lung nodules with CT at various dose levels. MATERIALS AND METHODS An anthropomorphic lung phantom and artificial lung nodules were used to simulate screening CT-examination at standard dose (100 mAs, 120 kVp) and 8 different low dose levels, using 120, 100 and 80 kVp combined with 100, 50 and 25 mAs. At each dose level 40 phantoms were randomly filled with 75 solid and 25 ground glass nodules (5-12 mm). Two radiologists and 3 different computer aided detection softwares (CAD) were paired to find the highest sensitivity. RESULTS Sensitivities at standard dose were 92%, 90%, 84%, 79% and 73% for reader 1, 2, CAD1, CAD2, CAD3, respectively. Combined sensitivity for human readers 1 and 2 improved to 97%, (p1=0.063, p2=0.016). Highest sensitivities--between 97% and 99.0%--were achieved by combining any radiologist with any CAD at any dose level. Combining any two CADs, sensitivities between 85% and 88% were significantly lower than for radiologists combined with CAD (p<0.03). CONCLUSIONS Combination of a human observer with any of the tested CAD systems provide optimal sensitivity for lung nodule detection even at reduced dose at 25 mAs/80 kVp.

Optimal dose levels in screening chest CT for unimpaired detection and volumetry of lung nodules, with and without computer assisted detection at minimal patient radiation

OBJECTIVES The aim of this phantom study was to minimize the radiation dose by finding the best combination of low tube current and low voltage that would result in accurate volume measurements when compared to standard CT imaging without significantly decreasing the sensitivity of detecting lung nodules both with and without the assistance of CAD. METHODS An anthropomorphic chest phantom containing artificial solid and ground glass nodules (GGNs, 5-12 mm) was examined with a 64-row multi-detector CT scanner with three tube currents of 100, 50 and 25 mAs in combination with three tube voltages of 120, 100 and 80 kVp. This resulted in eight different protocols that were then compared to standard CT sensitivity (100 mAs/120 kVp). For each protocol, at least 127 different nodules were scanned in 21-25 phantoms. The nodules were analyzed in two separate sessions by three independent, blinded radiologists and computer-aided detection (CAD) software. RESULTS The mean sensitivity of the radiologists for identifying solid lung nodules on a standard CT was 89.7% ± 4.9%. The sensitivity was not significantly impaired when the tube and current voltage were lowered at the same time, except at the lowest exposure level of 25 mAs/80 kVp [80.6% ± 4.3% (p = 0.031)]. Compared to the standard CT, the sensitivity for detecting GGNs was significantly lower at all dose levels when the voltage was 80 kVp; this result was independent of the tube current. The CAD significantly increased the radiologists' sensitivity for detecting solid nodules at all dose levels (5-11%). No significant volume measurement errors (VMEs) were documented for the radiologists or the CAD software at any dose level. CONCLUSIONS Our results suggest a CT protocol with 25 mAs and 100 kVp is optimal for detecting solid and ground glass nodules in lung cancer screening. The use of CAD software is highly recommended at all dose levels.

A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

In this paper, a computer-aided diagnostic (CAD) system for the classification of hepatic lesions from computed tomography (CT) images is presented. Regions of interest (ROIs) taken from nonenhanced CT images of normal liver, hepatic cysts, hemangiomas, and hepatocellular carcinomas have been used as input to the system. The proposed system consists of two modules: the feature extraction and the classification modules. The feature extraction module calculates the average gray level and 48 texture characteristics, which are derived from the spatial gray-level co-occurrence matrices, obtained from the ROIs. The classifier module consists of three sequentially placed feed-forward neural networks (NNs). The first NN classifies into normal or pathological liver regions. The pathological liver regions are characterized by the second NN as cyst or "other disease." The third NN classifies "other disease" into hemangioma or hepatocellular carcinoma. Three feature selection techniques have been applied to each individual NN: the sequential forward selection, the sequential floating forward selection, and a genetic algorithm for feature selection. The comparative study of the above dimensionality reduction methods shows that genetic algorithms result in lower dimension feature vectors and improved classification performance.

Computer-aided diagnosis of carotid atherosclerosis based on ultrasound image statistics, laws' texture and neural networks

Quantitative characterisation of carotid atherosclerosis and classification into symptomatic or asymptomatic is crucial in planning optimal treatment of atheromatous plaque. The computer-aided diagnosis (CAD) system described in this paper can analyse ultrasound (US) images of carotid artery and classify them into symptomatic or asymptomatic based on their echogenicity characteristics. The CAD system consists of three modules: a) the feature extraction module, where first-order statistical (FOS) features and Laws' texture energy can be estimated, b) the dimensionality reduction module, where the number of features can be reduced using analysis of variance (ANOVA), and c) the classifier module consisting of a neural network (NN) trained by a novel hybrid method based on genetic algorithms (GAs) along with the back propagation algorithm. The hybrid method is able to select the most robust features, to adjust automatically the NN architecture and to optimise the classification performance. The performance is measured by the accuracy, sensitivity, specificity and the area under the receiver-operating characteristic (ROC) curve. The CAD design and development is based on images from 54 symptomatic and 54 asymptomatic plaques. This study demonstrates the ability of a CAD system based on US image analysis and a hybrid trained NN to identify atheromatous plaques at high risk of stroke.

Improved detection of pulmonary nodules on energy-subtracted chest radiographs with a commercial computer-aided diagnosis software: comparison with human observers

To retrospectively analyze the performance of a commercial computer-aided diagnosis (CAD) software in the detection of pulmonary nodules in original and energy-subtracted (ES) chest radiographs.

Learning real-world stimuli in a neural network with spike-driven synaptic dynamics

We present a model of spike-driven synaptic plasticity inspired by experimental observations and motivated by the desire to build an electronic hardware device that can learn to classify complex stimuli in a semisupervised fashion. During training, patterns of activity are sequentially imposed on the input neurons, and an additional instructor signal drives the output neurons toward the desired activity. The network is made of integrate-and-fire neurons with constant leak and a floor. The synapses are bistable, and they are modified by the arrival of presynaptic spikes. The sign of the change is determined by both the depolarization and the state of a variable that integrates the postsynaptic action potentials. Following the training phase, the instructor signal is removed, and the output neurons are driven purely by the activity of the input neurons weighted by the plastic synapses. In the absence of stimulation, the synapses preserve their internal state indefinitely. Memories are also very robust to the disruptive action of spontaneous activity. A network of 2000 input neurons is shown to be able to classify correctly a large number (thousands) of highly overlapping patterns (300 classes of preprocessed Latex characters, 30 patterns per class, and a subset of the NIST characters data set) and to generalize with performances that are better than or comparable to those of artificial neural networks. Finally we show that the synaptic dynamics is compatible with many of the experimental observations on the induction of long-term modifications (spike-timing-dependent plasticity and its dependence on both the postsynaptic depolarization and the frequency of pre- and postsynaptic neurons).

Neural network based glucose - insulin metabolism models for children with Type 1 diabetes

In this paper two models for the simulation of glucose-insulin metabolism of children with Type 1 diabetes are presented. The models are based on the combined use of Compartmental Models (CMs) and artificial Neural Networks (NNs). Data from children with Type 1 diabetes, stored in a database, have been used as input to the models. The data are taken from four children with Type 1 diabetes and contain information about glucose levels taken from continuous glucose monitoring system, insulin intake and food intake, along with corresponding time. The influences of taken insulin on plasma insulin concentration, as well as the effect of food intake on glucose input into the blood from the gut, are estimated from the CMs. The outputs of CMs, along with previous glucose measurements, are fed to a NN, which provides short-term prediction of glucose values. For comparative reasons two different NN architectures have been tested: a Feed-Forward NN (FFNN) trained with the back-propagation algorithm with adaptive learning rate and momentum, and a Recurrent NN (RNN), trained with the Real Time Recurrent Learning (RTRL) algorithm. The results indicate that the best prediction performance can be achieved by the use of RNN.