scholarly journals Application of Deep Convolution Network to Automated Image Segmentation of Chest CT for Patients With Tumor

2021 ◽  
Vol 11 ◽  
Author(s):  
Hui Xie ◽  
Jian-Fang Zhang ◽  
Qing Li
Keyword(s):  
Image Segmentation ◽  
Network Model ◽  
Chest Ct ◽  
Automated Segmentation ◽  
Training Set ◽  
Test Set ◽  
Convolutional Networks ◽  
Optimal Segmentation ◽  
Validation Set ◽  
Axial Slice

ObjectivesTo automate image delineation of tissues and organs in oncological radiotherapy by combining the deep learning methods of fully convolutional network (FCN) and atrous convolution (AC).MethodsA total of 120 sets of chest CT images of patients were selected, on which radiologists had outlined the structures of normal organs. Of these 120 sets of images, 70 sets (8,512 axial slice images) were used as the training set, 30 sets (5,525 axial slice images) as the validation set, and 20 sets (3,602 axial slice images) as the test set. We selected 5 published FCN models and 1 published Unet model, and then combined FCN with AC algorithms to generate 3 improved deep convolutional networks, namely, dilation fully convolutional networks (D-FCN). The images in the training set were used to fine-tune and train the above 8 networks, respectively. The images in the validation set were used to validate the 8 networks in terms of the automated identification and delineation of organs, in order to obtain the optimal segmentation model of each network. Finally, the images of the test set were used to test the optimal segmentation models, and thus we evaluated the capability of each model of image segmentation by comparing their Dice coefficients between automated and physician delineation.ResultsAfter being fully tuned and trained with the images in the training set, all the networks in this study performed well in automated image segmentation. Among them, the improved D-FCN 4s network model yielded the best performance in automated segmentation in the testing experiment, with an global Dice of 87.11%, and a Dice of 87.11%, 97.22%, 97.16%, 89.92%, and 70.51% for left lung, right lung, pericardium, trachea, and esophagus, respectively.ConclusionWe proposed an improved D-FCN. Our results showed that this network model might effectively improve the accuracy of automated segmentation of the images in thoracic radiotherapy, and simultaneously perform automated segmentation of multiple targets.

scholarly journals Application of Multi-Scale Fusion Attention U-Net to Segment the Thyroid Gland on CT Localization Images for Radiotherapy

2021 ◽  
Author(s):  
Xiaobo Wen ◽  
Biao Zhao ◽  
Meifang Yuan ◽  
Jinzhi Li ◽  
Mengzhen Sun ◽  
...  
Keyword(s):  
Thyroid Gland ◽  
Clinical Work ◽  
Similarity Coefficient ◽  
Dice Similarity Coefficient ◽  
Training Set ◽  
Data Set ◽  
Test Set ◽  
Noise Interference ◽  
Multi Scale ◽  
Validation Set

Abstract Objectives: To explore the performance of Multi-scale Fusion Attention U-net (MSFA-U-net) in thyroid gland segmentation on CT localization images for radiotherapy. Methods: CT localization images for radiotherapy of 80 patients with breast cancer or head and neck tumors were selected; label images were manually delineated by experienced radiologists. The data set was randomly divided into the training set (n=60), the validation set (n=10), and the test set (n=10). Data expansion was performed in the training set, and the performance of the MSFA-U-net model was evaluated using the evaluation indicators Dice similarity coefficient (DSC), Jaccard similarity coefficient (JSC), positive predictive value (PPV), sensitivity (SE), and Hausdorff distance (HD). Results: With the MSFA-U-net model, the DSC, JSC, PPV, SE, and HD indexes of the segmented thyroid gland in the test set were 0.8967±0.0935, 0.8219±0.1115, 0.9065±0.0940, 0.8979±0.1104, and 2.3922±0.5423, respectively. Compared with U-net, HR-net, and Attention U-net, MSFA-U-net showed that DSC increased by 0.052, 0.0376, and 0.0346 respectively; JSC increased by 0.0569, 0.0805, and 0.0433, respectively; SE increased by 0.0361, 0.1091, and 0.0831, respectively; and HD increased by −0.208, −0.1952, and −0.0548, respectively. The test set image results showed that the thyroid edges segmented by the MSFA-U-net model were closer to the standard thyroid delineated by the experts, in comparison with those segmented by the other three models. Moreover, the edges were smoother, over-anti-noise interference was stronger, and oversegmentation and undersegmentation were reduced. Conclusion: The MSFA-U-net model can meet basic clinical requirements and improve the efficiency of physicians' clinical work.

scholarly journals Multiclass Classifier for P-Glycoprotein Substrates, Inhibitors, and Non-Active Compounds

Molecules ◽  
2019 ◽  
Vol 24 (10) ◽  
pp. 2006 ◽  
Author(s):  
Liadys Mora Lagares ◽  
Nikola Minovski ◽  
Marjana Novič
Keyword(s):  
In Silico ◽  
Transmembrane Protein ◽  
External Validation ◽  
Assessment Process ◽  
Classification Model ◽  
Training Set ◽  
Test Set ◽  
Active Compounds ◽  
P Glycoprotein ◽  
Validation Set

P-glycoprotein (P-gp) is a transmembrane protein that actively transports a wide variety of chemically diverse compounds out of the cell. It is highly associated with the ADMET (absorption, distribution, metabolism, excretion and toxicity) properties of drugs/drug candidates and contributes to decreasing toxicity by eliminating compounds from cells, thereby preventing intracellular accumulation. Therefore, in the drug discovery and toxicological assessment process it is advisable to pay attention to whether a compound under development could be transported by P-gp or not. In this study, an in silico multiclass classification model capable of predicting the probability of a compound to interact with P-gp was developed using a counter-propagation artificial neural network (CP ANN) based on a set of 2D molecular descriptors, as well as an extensive dataset of 2512 compounds (1178 P-gp inhibitors, 477 P-gp substrates and 857 P-gp non-active compounds). The model provided a good classification performance, producing non error rate (NER) values of 0.93 for the training set and 0.85 for the test set, while the average precision (AvPr) was 0.93 for the training set and 0.87 for the test set. An external validation set of 385 compounds was used to challenge the model’s performance. On the external validation set the NER and AvPr values were 0.70 for both indices. We believe that this in silico classifier could be effectively used as a reliable virtual screening tool for identifying potential P-gp ligands.

scholarly journals Dataset Splitting Techniques Comparison For Face Classification on CCTV Images

2020 ◽  
Vol 14 (4) ◽  
pp. 341
Author(s):  
Ade Nurhopipah ◽  
Uswatun Hasanah
Keyword(s):  
Splitting Method ◽  
Machine Learning Algorithms ◽  
Support Vector ◽  
Training Set ◽  
Test Set ◽  
Face Classification ◽  
Lower Accuracy ◽  
Svm Algorithm ◽  
Stable Performance ◽  
Validation Set

The performance of classification models in machine learning algorithms is influenced by many factors, one of which is dataset splitting method. To avoid overfitting, it is important to apply a suitable dataset splitting strategy. This study presents comparison of four dataset splitting techniques, namely Random Sub-sampling Validation (RSV), k-Fold Cross Validation (k-FCV), Bootstrap Validation (BV) and Moralis Lima Martin Validation (MLMV). This comparison is done in face classification on CCTV images using Convolutional Neural Network (CNN) algorithm and Support Vector Machine (SVM) algorithm. This study is also applied in two image datasets. The results of the comparison are reviewed by using model accuracy in training set, validation set and test set, also bias and variance of the model. The experiment shows that k-FCV technique has more stable performance and provide high accuracy on training set as well as good generalizations on validation set and test set. Meanwhile, data splitting using MLMV technique has lower performance than the other three techniques since it yields lower accuracy. This technique also shows higher bias and variance values and it builds overfitting models, especially when it is applied on validation set.

scholarly journals An Early Biomarker Algorithm Predicts Lethal Graft-Versus-Host Disease and Survival after Allogeneic Hematopoietic Cell Transplantation

Blood ◽  
2016 ◽  
Vol 128 (22) ◽  
pp. 509-509 ◽  
Author(s):  
Matthew J Hartwell ◽  
Umut Ozbek ◽  
Ernst Holler ◽  
Anne S. Renteria ◽  
Pavan R. Reddy ◽  
...  
Keyword(s):  
High Risk ◽  
Research Funding ◽  
Acute Gvhd ◽  
Conditioning Regimen ◽  
Training Set ◽  
Test Set ◽  
Graft Versus Host ◽  
Independent Validation ◽  
Blood Biomarker ◽  
Validation Set

Abstract No laboratory test can predict non-relapse mortality (NRM) after hematopoietic cellular transplantation (HCT) prior to the onset graft-versus-host disease (GVHD). Recently, we have shown that a signature of three GVHD plasma biomarkers (TNFR1, ST2, and REG3α) can predict response to GVHD therapy and NRM at the onset of clinical GVHD (Levine, Lancet Haem, 2015). Our goal in the current study was to identify a blood biomarker signature that could predict lethal GVHD and six-month NRM well in advance of the onset of GVHD symptoms. Patient samples on day +7 after HCT were obtained from 1,287 patients from 11 HCT centers in the Mount Sinai Acute GVHD International Consortium (MAGIC). Samples from two large centers (n = 929) were combined and randomly assigned to a training set (n = 620) and test set (n = 309). 358 patients from nine others centers constituted an independent validation set. The overall cumulative incidences of 6-month NRM were 11%, 12%, and 13% for the training, test, and validation sets respectively. The incidence of lethal GVHD, defined as death without preceding relapse while under steroid treatment for acute GVHD, were 18%, 24%, and 14% in the same groups, respectively. The median day of GVHD onset was 28 days in the training set and 29 days in the test and validation sets. We measured four GVHD related biomarkers [ST2, REG3α, TNFR1, and IL2Rα] in all samples and used the training set alone to develop competing risks regression models that used all 13 possible combinations of one to four biomarkers to predict 6-month NRM. The best algorithm, which we rigorously confirmed through Monte Carlo cross-validation of 75 different combinations of training sets, included ST2 and REG3α. No combination of one, three, or four biomarkers was superior to the combination of these two biomarkers. The day 7 algorithm identified high risk (HR) and low risk (LR) groups with 6-month NRMs of 28% and 7%, respectively (p<0.001) (Fig 1A). The relapse rates did not differ between risk groups so that overall survival (OS) was 60% for HR and 84% for LR (p<0.001) (Fig 1B). When applied to the test set (Fig 1C/D), the algorithm identified 54/309 (17%) of the patients as HR with an NRM of 33% vs 7% for LR patients (p<0.001) and 6-month OS of 57% and 81% for HR and LR patients, respectively (p<0.001). In the independent validation set (Fig 1 E/F), the algorithm identified 72/358 (20%) of the patients as HR with an NRM of 26% vs 10% for LR patients (p<0.001) and OS of 68% and 85% for HR and LR patients, respectively (p<0.001). High risk patients were three times more likely to die from GVHD than LR patients in each cohort (p<0.001) (Fig 2). The GI tract is the GVHD target organ that is most resistant to treatment and represents a major cause of NRM, and we observed twice as much severe GI GVHD (stage 3 or 4) in HR patients as in LR patients (p<0.001, data not shown). The algorithm successfully separated HR and LR strata for 6 month NRM in several groups with differing risks for GVHD and NRM, including donor type, degree. of HLA-match, age group, and conditioning regimen intensity (Fig 3). In conclusion, we have developed a blood biomarker algorithm that predicts the development of lethal GVHD seven days after HCT, which performed successfully in large multicenter validation sets. The GVH reaction is already in progress by day +7, even though clinical symptoms may not occur until days or weeks later. We speculate that the blood biomarker concentrations at this early time point reflect subclinical GI pathology, a notion that is reinforced by the fact that ST2 and REG3α, the two biomarkers in the algorithm, are closely associated with GI GVHD. The algorithm identified HR and LR strata in several patient groups with different overall risk for lethal GVHD (donor, HLA match, conditioning regimen intensity, age). This day +7 algorithm should prove useful in clinical BMT research by identifying patients at high risk for lethal GVHD who might benefit from aggressive preemptive treatment strategies. Disclosures Chen: Novartis: Research Funding; Incyte Corporation: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding. Jagasia:Therakos: Consultancy. Kitko:Therakos: Honoraria, Speakers Bureau. Kroeger:Novartis: Honoraria, Research Funding. Levine:Viracor: Patents & Royalties: GVHD biomarkers patent. Ferrara:Viracor: Patents & Royalties: GVHD biomarkers patent.

scholarly journals Identification of 5 Gene Signatures in Survival Prediction for Patients with Lung Squamous Cell Carcinoma Based on Integrated Multiomics Data Analysis

2020 ◽  
Vol 2020 ◽  
pp. 1-19
Author(s):  
Hongxia Ma ◽  
Lihong Tong ◽  
Qian Zhang ◽  
Wenjun Chang ◽  
Fengsen Li
Keyword(s):  
Squamous Cell Carcinoma ◽  
Data Analysis ◽  
Cell Carcinoma ◽  
Squamous Cell ◽  
Lung Squamous Cell Carcinoma ◽  
Training Set ◽  
Test Set ◽  
Gene Signatures ◽  
Genomic Variants ◽  
Validation Set

Background. Lung squamous cell carcinoma (LSCC) is a frequently diagnosed cancer worldwide, and it has a poor prognosis. The current study is aimed at developing the prediction of LSCC prognosis by integrating multiomics data including transcriptome, copy number variation data, and mutation data analysis, so as to predict patients’ survival and discover new therapeutic targets. Methods. RNASeq, SNP, CNV data, and LSCC patients’ clinical follow-up information were downloaded from The Cancer Genome Atlas (TCGA), and the samples were randomly divided into two groups, namely, the training set and the validation set. In the training set, the genes related to prognosis and those with different copy numbers or with different SNPs were integrated to extract features using random forests, and finally, robust biomarkers were screened. In addition, a gene-related prognostic model was established and further verified in the test set and GEO validation set. Results. We obtained a total of 804 prognostic-related genes and 535 copy amplification genes, 621 copy deletions genes, and 388 significantly mutated genes in genomic variants; noticeably, these genomic variant genes were found closely related to tumor development. A total of 51 candidate genes were obtained by integrating genomic variants and prognostic genes, and 5 characteristic genes (HIST1H2BH, SERPIND1, COL22A1, LCE3C, and ADAMTS17) were screened through random forest feature selection; we found that many of those genes had been reported to be related to LSCC progression. Cox regression analysis was performed to establish 5-gene signature that could serve as an independent prognostic factor for LSCC patients and can stratify risk samples in training set, test set, and external validation set (p<0.01), and the 5-year survival areas under the curve (AUC) of both training set and validation set were > 0.67. Conclusion. In the current study, 5 gene signatures were constructed as novel prognostic markers to predict the survival of LSCC patients. The present findings provide new diagnostic and prognostic biomarkers and therapeutic targets for LSCC treatment.

scholarly journals Discrimination Between Invasive and In Situ Melanomas Using Clinical Close-Up Images and a De Novo Convolutional Neural Network

2021 ◽  
Vol 8 ◽  
Author(s):  
Sam Polesie ◽  
Martin Gillstedt ◽  
Gustav Ahlgren ◽  
Hannah Ceder ◽  
Johan Dahlén Gyllencreutz ◽  
...  
Keyword(s):  
De Novo ◽  
Area Under The Curve ◽  
Classification Problem ◽  
Dense Layer ◽  
Training Set ◽  
Clinical Images ◽  
Test Set ◽  
Time Period ◽  
Validation Set

Background: Melanomas are often easy to recognize clinically but determining whether a melanoma is in situ (MIS) or invasive is often more challenging even with the aid of dermoscopy. Recently, convolutional neural networks (CNNs) have made significant and rapid advances within dermatology image analysis. The aims of this investigation were to create a de novo CNN for differentiating between MIS and invasive melanomas based on clinical close-up images and to compare its performance on a test set to seven dermatologists.Methods: A retrospective study including clinical images of MIS and invasive melanomas obtained from our department during a five-year time period (2016–2020) was conducted. Overall, 1,551 images [819 MIS (52.8%) and 732 invasive melanomas (47.2%)] were available. The images were randomized into three groups: training set (n = 1,051), validation set (n = 200), and test set (n = 300). A de novo CNN model with seven convolutional layers and a single dense layer was developed.Results: The area under the curve was 0.72 for the CNN (95% CI 0.66–0.78) and 0.81 for dermatologists (95% CI 0.76–0.86) (P &lt; 0.001). The CNN correctly classified 208 out of 300 lesions (69.3%) whereas the corresponding number for dermatologists was 216 (72.0%). When comparing the CNN performance to each individual reader, three dermatologists significantly outperformed the CNN.Conclusions: For this classification problem, the CNN was outperformed by the dermatologist. However, since the algorithm was only trained and validated on 1,251 images, future refinement and development could make it useful for dermatologists in a real-world setting.

A 13-immune gene set signature for prediction of colon cancer prognosis

2020 ◽  
Vol 23 ◽  
Author(s):  
Zhuo Lu ◽  
Jin Chen ◽  
JiongYi Yan ◽  
QiaoMing Liu ◽  
Fang Li ◽  
...  
Keyword(s):  
Colon Cancer ◽  
External Validation ◽  
Therapeutic Targets ◽  
The Cancer Genome Atlas ◽  
Cancer Prognosis ◽  
Immune Gene ◽  
Training Set ◽  
Test Set ◽  
Gene Set ◽  
Validation Set

Background: Colon cancer is one of the most common cancer worldwide and has a poor prognosis. Through the analysis of transcriptome and clinical data of colon cancer, immune gene-set signature was identified by single sample enrichment analysis (ssGSEA) scoring to predict patient survival and discover new therapeutic targets. Objective: To study the role of immune gene-set signature in colon cancer. Methods: First, RNASeq and clinical follow-up information were downloaded from The Cancer Genome Atlas (TCGA). Immune gene-related gene sets were collected from ImmPort database. Genes and immunological pathways related to prognosis were screened in the training set and integrated for feature selection using random forest. Immune gene-related prognosis model was verified in the entire TCGA test set and GEO validation set and compared with immune cells scores and matrix score. Results: 1650 prognostic genes and 13 immunological pathways were identified. These genes and pathways are closely related to the development of tumors. 13-immune gene-set signature was established, which is an independent prognostic factor for patients with colon cancer. Risk stratification of samples could be carried out in the training set, test set and external validation set. The AUC of five-year survival in the training set and validation set is greater than 0.6. Immunosuppression occurs in high-risk samples. Compared with published models, Riskscore has better prediction effect. Conclusion: This study constructed 13-immune gene-set signature as a new prognostic marker to predict the survival of patients with colon cancer, and provided new diagnostic/prognostic biomarkers and therapeutic targets for colon cancer.

scholarly journals Diagnostic Classification of Cystoscopic Images Using Deep Convolutional Neural Networks

2018 ◽  
pp. 1-8 ◽  
Author(s):  
Okyaz Eminaga ◽  
Nurettin Eminaga ◽  
Axel Semjonow ◽  
Bernhard Breil
Keyword(s):  
Deep Learning ◽  
Harmonic Series ◽  
Diagnostic Classification ◽  
Training Set ◽  
Deep Convolutional Neural Networks ◽  
Data Set ◽  
Test Set ◽  
Filter Size ◽  
Validation Set

Purpose The recognition of cystoscopic findings remains challenging for young colleagues and depends on the examiner’s skills. Computer-aided diagnosis tools using feature extraction and deep learning show promise as instruments to perform diagnostic classification. Materials and Methods Our study considered 479 patient cases that represented 44 urologic findings. Image color was linearly normalized and was equalized by applying contrast-limited adaptive histogram equalization. Because these findings can be viewed via cystoscopy from every possible angle and side, we ultimately generated images rotated in 10-degree grades and flipped them vertically or horizontally, which resulted in 18,681 images. After image preprocessing, we developed deep convolutional neural network (CNN) models (ResNet50, VGG-19, VGG-16, InceptionV3, and Xception) and evaluated these models using F1 scores. Furthermore, we proposed two CNN concepts: 90%-previous-layer filter size and harmonic-series filter size. A training set (60%), a validation set (10%), and a test set (30%) were randomly generated from the study data set. All models were trained on the training set, validated on the validation set, and evaluated on the test set. Results The Xception-based model achieved the highest F1 score (99.52%), followed by models that were based on ResNet50 (99.48%) and the harmonic-series concept (99.45%). All images with cancer lesions were correctly determined by these models. When the focus was on the images misclassified by the model with the best performance, 7.86% of images that showed bladder stones with indwelling catheter and 1.43% of images that showed bladder diverticulum were falsely classified. Conclusion The results of this study show the potential of deep learning for the diagnostic classification of cystoscopic images. Future work will focus on integration of artificial intelligence–aided cystoscopy into clinical routines and possibly expansion to other clinical endoscopy applications.

scholarly journals A machine learning model for screening of body fluid cytology smears

2021 ◽  
Author(s):  
Parikshit Sanyal ◽  
Sayak Paul ◽  
Vandana Rana ◽  
Kanchan Kulhari
Keyword(s):  
Machine Learning ◽  
Body Fluid ◽  
Learning Model ◽  
Malignant Cells ◽  
Training Set ◽  
Test Set ◽  
White Balance ◽  
Machine Learning Model ◽  
Validation Set ◽  
Fluid Cytology

Introduction: Body fluid cytology is one of the commonest investigations performed in indoor patients, both for diagnosis of suspected carcinoma as well as staging of known carcinoma. Carcinoma is diagnosed in body fluids by the pathologist through microscopic examination and searching for malignant epithelial cell clusters. The process of screening body fluid smears is a time consuming and error prone process. Aim: We have attempted to construct a machine learning model which can screen body fluid cytology smears for malignant cells. Materials and methods: MGG stained Ascitic / pleural fluid cytology smears were included from 21 cases (14 malignant, 07 benign) in this study. A total of 693 microphotographs were taken at 40x magnification at the same illumination and after correction of white balance. A Magnus Microphotography system was used for photography. The images were split into the training set (195 images), test set (120 images) and validation set (378 images). A machine learning model, a convolutional neural network, was developed in the Python programming language using the Keras deep learning library. The model was trained with the images of the training set. After completion of training, the model was evaluated on the test set of images. Results: Evaluation of the model on the test set produced a sensitivity of 97.87%, specificity 85.26%, PPV 95.18%, NPV 93.10% In 06 images, the model has failed to detect singly scattered malignant cells/ clusters. 14 (3.7%) false positives was reported by the model. The machine learning model shows potential utility as a screening tool. However, it needs improvement in detecting singly scattered malignant cells and filtering inflammatory infiltrate.

scholarly journals Synthesizing the First Phase of Dynamic Sequences of Breast MRI for Enhanced Lesion Identification

2021 ◽  
Vol 11 ◽  
Author(s):  
Pingping Wang ◽  
Pin Nie ◽  
Yanli Dang ◽  
Lifang Wang ◽  
Kaiguo Zhu ◽  
...  
Keyword(s):  
Predictive Value ◽  
Signal To Noise Ratio ◽  
Breast Mri ◽  
Structural Similarity ◽  
Image Synthesis ◽  
Absolute Error ◽  
Diagnostic Value ◽  
Training Set ◽  
Test Set ◽  
Validation Set

ObjectiveTo develop a deep learning model for synthesizing the first phases of dynamic (FP-Dyn) sequences to supplement the lack of information in unenhanced breast MRI examinations.MethodsIn total, 97 patients with breast MRI images were collected as the training set (n = 45), the validation set (n = 31), and the test set (n = 21), respectively. An enhance border lifelike synthesize (EDLS) model was developed in the training set and used to synthesize the FP-Dyn images from the T1WI images in the validation set. The peak signal-to-noise ratio (PSNR), structural similarity (SSIM), mean square error (MSE) and mean absolute error (MAE) of the synthesized images were measured. Moreover, three radiologists subjectively assessed image quality, respectively. The diagnostic value of the synthesized FP-Dyn sequences was further evaluated in the test set.ResultsThe image synthesis performance in the EDLS model was superior to that in conventional models from the results of PSNR, SSIM, MSE, and MAE. Subjective results displayed a remarkable visual consistency between the synthesized and original FP-Dyn images. Moreover, by using a combination of synthesized FP-Dyn sequence and an unenhanced protocol, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MRI were 100%, 72.73%, 76.92%, and 100%, respectively, which had a similar diagnostic value to full MRI protocols.ConclusionsThe EDLS model could synthesize the realistic FP-Dyn sequence to supplement the lack of enhanced images. Compared with full MRI examinations, it thus provides a new approach for reducing examination time and cost, and avoids the use of contrast agents without influencing diagnostic accuracy.

Sign in / Sign up

Export Citation Format

Share Document