A Segmentation Framework of Pulmonary Nodules in Lung CT Images

Mukhopadhyay, Sudipta

doi:10.1007/s10278-015-9801-9

Cited by 66 publications

(40 citation statements)

References 24 publications

(41 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our model achieves the highest DSC score of 0.8483 and it outperforms existing methods. The traditional segmentation techniques by Mukhopadhyay cannot adapt to large variation of nodules such as size, shape, and texture. Although both methods by Çiçek et al .…”

Section: Experiments and Resultsmentioning

confidence: 86%

“…The comparison of segmentation results with state‐of‐the‐art methods is given in Table . All the methods are evaluated on LIDC‐IDRI dataset and the commonly used metric is Dice coefficient.…”

Section: Experiments and Resultsmentioning

confidence: 99%

“…Table provided comparison with state‐of‐the‐art methods . The performance of the traditional method by Mukhopadhyay is poor because it requires careful tuning of hyperparameters (e.g., thresholding value of density for different nodules), which triggers off weak generalization ability on large dataset. Although Çiçek et al .…”

Section: Discussionmentioning

confidence: 99%

“…Few methods were developed for segmentation of all solid, part‐solid, and GGO nodules . Kubota et al .…”

Section: Introductionmentioning

confidence: 99%

“…Nodules are automatically segmented with an improved shape break‐and‐repair strategy. Mukhopadhyay adopted a two‐step segmentation method. It first categorizes nodules by internal texture.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Pulmonary nodule segmentation with CT sample synthesis using adversarial networks

et al. 2019

View full text Add to dashboard Cite

Purpose Segmentation of pulmonary nodules is critical for the analysis of nodules and lung cancer diagnosis. We present a novel framework of segmentation for various types of nodules using convolutional neural networks (CNNs). Methods The proposed framework is composed of two major parts. The first part is to increase the variety of samples and build a more balanced dataset. A conditional generative adversarial network (cGAN) is employed to produce synthetic CT images. Semantic labels are generated to impart spatial contextual knowledge to the network. Nine attribute scoring labels are combined as well to preserve nodule features. To refine the realism of synthesized samples, reconstruction error loss is introduced into cGAN. The second part is to train a nodule segmentation network on the extended dataset. We build a three‐dimensional (3D) CNN model that exploits heterogeneous maps including edge maps and local binary pattern maps. The incorporation of these maps informs the model of texture patterns and boundary information of nodules, which assists high‐level feature learning for segmentation. Residual unit, which learns to reduce residual error, is adopted to accelerate training and improve accuracy. Results Validation on LIDC‐IDRI dataset demonstrates that the generated samples are realistic. The mean squared error and average cosine similarity between real and synthesized samples are 1.55×10−2 and 0.9534, respectively. The Dice coefficient, positive predicted value, sensitivity, and accuracy are, respectively, 0.8483, 0.8895, 0.8511, and 0.9904 for the segmentation results. Conclusions The proposed 3D CNN segmentation framework, based on the use of synthesized samples and multiple maps with residual learning, achieves more accurate nodule segmentation compared to existing state‐of‐the‐art methods. The proposed CT image synthesis method can not only output samples close to real images but also allow for stochastic variation in image diversity.

show abstract

Section: Experiments and Resultsmentioning

confidence: 86%

Section: Experiments and Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

“…Few methods were developed for segmentation of all solid, part‐solid, and GGO nodules . Kubota et al .…”

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Pulmonary nodule segmentation with CT sample synthesis using adversarial networks

et al. 2019

View full text Add to dashboard Cite

show abstract

Radiomics Models Based on Apparent Diffusion Coefficient Maps for the Prediction of High‐Grade Prostate Cancer at Radical Prostatectomy: Comparison With Preoperative Biopsy

Han

Liu

et al. 2021

Magnetic Resonance Imaging

View full text Add to dashboard Cite

Background It is feasible to use magnetic resonance (MR)‐based radiomics to distinguish high‐grade from low‐grade prostate cancer (PCa), but radiomics model performance based on fully automated segmentation remains unknown. Purpose To develop and test radiomics models based on manually or automatically gained masks on apparent diffusion coefficient (ADC) maps to predict high‐grade (Gleason score ≥ 4 + 3) PCa at radical prostatectomy (RP). Study Type Retrospective. Population A total of 176 patients (94 high‐grade PCa and 82 low‐grade PCa) with complete RP, preoperative biopsy, and multiparametric magnetic resonance imaging (mpMRI) were retrospectively recruited and randomly divided into training (N = 123) and test (N = 53) cohorts. Field Strength/Sequence Using a 3.0‐T MR scanner, ADC maps were calculated from diffusion‐weighted imaging (b values = 0, 1400 s/mm2, echo planar imaging). Assessment Two radiologists segmented the whole prostate gland and the most index prostate lesion. Automatic segmentation of the prostate and the lesion were performed. Four radiomics models were constructed using four masks (manual/automatic prostate gland/PCa lesion segmentation). According to the standard reference of the RP histopathologic assessment, the performance of each radiomics models was compared with that of biopsy and Prostate Imaging Reporting and Data System version 2.1 (PI‐RADS) assessment. Statistical Tests A receiver operating characteristic curve analysis was employed to estimate the area under the curve (AUC) values of the models. The AUCs of the four models, biopsy, and PI‐RADS assessment were compared using the DeLong test. Results The four radiomics models yielded AUCs of 0.710, 0.731, 0.726, and 0.709 in the test cohort, respectively; biopsy and PI‐RADS assessment yielded AUCs of 0.793 and 0.680, respectively. No significant differences were found among model, biopsy, and PI‐RADS assessment comparisons (P = 0.132–0.988). Data Conclusion To distinguish high‐grade from low‐grade PCa, radiomics models based on automatic segmentation on ADC maps exhibit approximately the same diagnostic efficacy as manual segmentation and biopsy, highlighting the possibility of a fully automatic workflow combining automated segmentation with radiomics analysis. Evidence Level 4 Technical Efficacy Stage 2

show abstract

Machine learning approach for distinguishing malignant and benign lung nodules utilizing standardized perinodular parenchymal features from CT

et al. 2019

View full text Add to dashboard Cite

Purpose: Computed tomography (CT) is an effective method for detecting and characterizing lung nodules in vivo. With the growing use of chest CT, the detection frequency of lung nodules is increasing. Noninvasive methods to distinguish malignant from benign nodules have the potential to decrease the clinical burden, risk, and cost involved in follow-up procedures on the large number of false-positive lesions detected. This study examined the benefit of including perinodular parenchymal features in machine learning (ML) tools for pulmonary nodule assessment. Methods: Lung nodule cases with pathology confirmed diagnosis (74 malignant, 289 benign) were used to extract quantitative imaging characteristics from computed tomography scans of the nodule and perinodular parenchyma tissue. A ML tool development pipeline was employed using k-medoids clustering and information theory to determine efficient predictor sets for different amounts of parenchyma inclusion and build an artificial neural network classifier. The resulting ML tool was validated using an independent cohort (50 malignant, 50 benign). Results: The inclusion of parenchymal imaging features improved the performance of the ML tool over exclusively nodular features (P < 0.01). The best performing ML tool included features derived from nodule diameter-based surrounding parenchyma tissue quartile bands. We demonstrate similar high-performance values on the independent validation cohort (AUC-ROC = 0.965). A comparison using the independent validation cohort with the Fleischner pulmonary nodule follow-up guidelines demonstrated a theoretical reduction in recommended follow-up imaging and procedures. Conclusions: Radiomic features extracted from the parenchyma surrounding lung nodules contain valid signals with spatial relevance for the task of lung cancer risk classification. Through standardization of feature extraction regions from the parenchyma, ML tool validation performance of 100% sensitivity and 96% specificity was achieved.

show abstract

A Segmentation Framework of Pulmonary Nodules in Lung CT Images

Cited by 66 publications

References 24 publications

Pulmonary nodule segmentation with CT sample synthesis using adversarial networks

Pulmonary nodule segmentation with CT sample synthesis using adversarial networks

Radiomics Models Based on Apparent Diffusion Coefficient Maps for the Prediction of High‐Grade Prostate Cancer at Radical Prostatectomy: Comparison With Preoperative Biopsy

Machine learning approach for distinguishing malignant and benign lung nodules utilizing standardized perinodular parenchymal features from CT

Contact Info

Product

Resources

About