Quantitative Computed Tomography Classification of Lung Nodules: Initial Comparison of 2- and 3-Dimensional Analysis.
ABSTRACT: The aim of this study was to compare the performance of 2- (2D) and 3-dimensional (3D) quantitative computed tomography (CT) methods for classifying lung nodules as lung cancer, metastases, or benign.Using semiautomated software and computerized analysis, we analyzed more than 50 quantitative CT features of 96 solid nodules in 94 patients, in 2D from a single slice and in 3D from the entire nodule volume. Multivariable logistic regression was used to classify nodule types. Model performance was assessed by the area under the receiver operating characteristic curve (AUC) using leave-one-out cross-validation.The AUC for distinguishing 53 primary lung cancers from 18 benign nodules and 25 metastases ranged from 0.79 to 0.83 and was not significantly different for 2D and 3D analyses (P = 0.29-0.78). Models distinguishing metastases from benign nodules were statistically significant only by 3D analysis (AUC = 0.84).Three-dimensional CT methods did not improve discrimination of lung cancer, but may help distinguish benign nodules from metastases.
Project description:Low-dose CT (LDCT) is widely accepted as the preferred method for detecting pulmonary nodules. However, the determination of whether a nodule is benign or malignant involves either repeated scans or invasive procedures that sample the lung tissue. Noninvasive methods to assess these nodules are needed to reduce unnecessary invasive tests. In this study, we have developed a pulmonary nodule classifier (PNC) using RNA from whole blood collected in RNA-stabilizing PAXgene tubes that addresses this need. Samples were prospectively collected from high-risk and incidental subjects with a positive lung CT scan. A total of 821 samples from 5 clinical sites were analyzed. Malignant samples were predominantly stage 1 by pathologic diagnosis and 97% of the benign samples were confirmed by 4 years of follow-up. A panel of diagnostic biomarkers was selected from a subset of the samples assayed on Illumina microarrays that achieved a ROC-AUC of 0.847 on independent validation. The microarray data were then used to design a biomarker panel of 559 gene probes to be validated on the clinically tested NanoString nCounter platform. RNA from 583 patients was used to assess and refine the NanoString PNC (nPNC), which was then validated on 158 independent samples (ROC-AUC = 0.825). The nPNC outperformed three clinical algorithms in discriminating malignant from benign pulmonary nodules ranging from 6-20 mm using just 41 diagnostic biomarkers. Overall, this platform provides an accurate, noninvasive method for the diagnosis of pulmonary nodules in patients with non-small cell lung cancer. SIGNIFICANCE: These findings describe a minimally invasive and clinically practical pulmonary nodule classifier that has good diagnostic ability at distinguishing benign from malignant pulmonary nodules.
Project description:<h4>Background</h4>The purpose of this study was to analyze if the use of texture analysis on spectral detector CT (SDCT)-derived iodine maps (IM) in addition to conventional images (CI) improves lung nodule differentiation, when being applied to a k-nearest neighbor (KNN) classifier.<h4>Methods</h4>183 cancer patients who underwent contrast-enhanced, venous phase SDCT of the chest were included: 85 patients with 146 benign lung nodules (BLN) confirmed by either prior/follow-up CT or histopathology and 98 patients with 425 lung metastases (LM) verified by histopathology, <sup>18</sup>F-FDG-PET-CT or unequivocal change during treatment. Semi-automatic 3D segmentation of BLN/LM was performed, and volumetric HU attenuation and iodine concentration were acquired. For conventional images and iodine maps, average, standard deviation, entropy, kurtosis, mean of the positive pixels (MPP), skewness, uniformity and uniformity of the positive pixels (UPP) within the volumes of interests were calculated. All acquired parameters were transferred to a KNN classifier.<h4>Results</h4>Differentiation between BLN and LM was most accurate, when using all CI-derived features combined with the most significant IM-derived feature, entropy (Accuracy:0.87; F1/Dice:0.92). However, differentiation accuracy based on the 4 most powerful CI-derived features performed only slightly inferior (Accuracy:0.84; F1/Dice:0.89, p=0.125). Mono-parametric lung nodule differentiation based on either feature alone (i.e. attenuation or iodine concentration) was poor (AUC=0.65, 0.58, respectively).<h4>Conclusions</h4>First-order texture feature analysis of contrast-enhanced staging SDCT scans of the chest yield accurate differentiation between benign and metastatic lung nodules. In our study cohort, the most powerful iodine map-derived feature slightly, yet insignificantly increased classification accuracy compared to classification based on conventional image features only.
Project description:OBJECTIVES:The post-imaging, mathematical predictive model was established by combining demographic and imaging characteristics with a pulmonary nodule risk score. The prediction model provides directions for the treatment of pulmonary nodules. Many studies have established predictive models for pulmonary nodules in different populations. However, the predictive factors contained in each model were significantly different. We hypothesized that applying different models to local research groups will make a difference in predicting the benign and malignant lung nodules, distinguishing between early and late lung cancers, and between adenocarcinoma and squamous cell carcinoma. In the present study, we compared four widely used and well-known mathematical prediction models. MATERIALS AND METHODS:We performed a retrospective study of 496 patients from January 2017 to October 2019, they were diagnosed with nodules by pathological. We evaluate models' performance by viewing 425 malignant and 71 benign patients' computed tomography results. At the same time, we use the calibration curve and the area under the receiver operating characteristic curve whose abbreviation is AUC to assess one model's predictive performance. RESULTS:We find that in distinguishing the Benign and the Malignancy, Peking University People's Hospital model possessed excellent performance (AUC = 0.63), as well as differentiating between early and late lung cancers (AUC = 0.67) and identifying lung adenocarcinoma (AUC = 0.61). While in the identification of lung squamous cell carcinoma, the Veterans Affairs model performed the best (AUC = 0.69). CONCLUSIONS:Geographic disparities are an extremely important influence factors, and which clinical features contained in the mathematical prediction model are the key to affect the precision and accuracy.
Project description:Accurately detecting and examining lung nodules early is key in diagnosing lung cancers and thus one of the best ways to prevent lung cancer deaths. Radiologists spend countless hours detecting small spherical-shaped nodules in computed tomography (CT) images. In addition, even after detecting nodule candidates, a considerable amount of effort and time is required for them to determine whether they are real nodules. The aim of this paper is to introduce a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules. In this paper, we use a three dimensional deep convolutional neural network (3D DCNN) with shortcut connections and a 3D DCNN with dense connections for lung nodule classification. The shortcut connections and dense connections successfully alleviate the gradient vanishing problem by allowing the gradient to pass quickly and directly. Connections help deep structured networks to obtain general as well as distinctive features of lung nodules. Moreover, we increased the dimension of DCNNs from two to three to capture 3D features. Compared with shallow 3D CNNs used in previous studies, deep 3D CNNs more effectively capture the features of spherical-shaped nodules. In addition, we use an alternative ensemble method called the checkpoint ensemble method to boost performance. The performance of our nodule classification method is compared with that of the state-of-the-art methods which were used in the LUng Nodule Analysis 2016 Challenge. Our method achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning. In the experimental setup ESB-ALL, the 3D DCNN with shortcut connections and the 3D DCNN with dense connections using the checkpoint ensemble method achieved the highest CPM score of 0.910. The result demonstrates that our method of using a 3D DCNN with shortcut connections, a 3D DCNN with dense connections, and the checkpoint ensemble method is effective for capturing 3D features of nodules and distinguishing nodules between non-nodules.
Project description: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.
Project description:<h4>Background</h4>The number of pulmonary nodules detected in the US is expected to increase substantially following recent recommendations for nationwide CT-based lung cancer screening. Given the low specificity of CT screening, non-invasive adjuvant methods are needed to differentiate cancerous lesions from benign nodules to help avoid unnecessary invasive procedures in the asymptomatic population. We have constructed a serum-based multi-biomarker panel and assessed its clinical accuracy in a retrospective analysis of samples collected from participants with suspicious radiographic findings in the Prostate, Lung, Chest and Ovarian (PLCO) cancer screening trial.<h4>Methods</h4>Starting with a set of 9 candidate biomarkers, we identified 8 that exhibited limited pre-analytical variability with increasing clotting time, a key pre-analytical variable associated with the collection of serum. These 8 biomarkers were evaluated in a training study consisting of 95 stage I NSCLC patients and 186 smoker controls where a 5-biomarker pulmonary nodule classifier (PNC) was selected. The clinical accuracy of the PNC was determined in a blinded study of asymptomatic individuals comprising 119 confirmed malignant nodule cases and 119 benign nodule controls selected from the PLCO screening trial.<h4>Results</h4>A PNC comprising 5 biomarkers: CEA, CYFRA 21-1, OPN, SCC, and TFPI, was selected in the training study. In an independent validation study, the PNC resolved lung cancer cases from benign nodule controls with an AUC of 0.653 (p < 0.0001). CEA and CYFRA 21-1, two of the markers included in the PNC, also accurately distinguished malignant lesions from benign controls.<h4>Conclusions</h4>A 5-biomarker blood test has been developed for the diagnostic evaluation of asymptomatic individuals with solitary pulmonary nodules.
Project description:Our objective was to compare the diagnostic performance and diagnostic confidence of convolutional neural networks (CNN) to radiologists in characterizing small hypoattenuating hepatic nodules (SHHN) in colorectal carcinoma (CRC) on CT scans. Retrospective review of CRC CT scans over 6-years yielded 199 patients (550 SHHN) defined as?<?1 cm in diameter. The reference standard was established through 1-year stability/MRI for benign or nodule evolution for malignant nodules. Five CNNs underwent supervised training on 150 patients (412 SHHN). The remaining 49 patients (138 SHHN) were used as testing-set to compare performance of 3 radiologists to CNN, measured through ROC AUC analysis of confidence rating assigned to each nodule by the radiologists. Multivariable modeling was used to compensate for radiologist bias from visible findings other than SHHN. In characterizing SHHN as benign or malignant, the radiologists' mean AUC ROC (0.96) was significantly higher than CNN (0.84, p?=?0.0004) but equivalent to CNN adjusted through multivariable modeling for presence of synchronous???1 cm liver metastases (0.95, p?=?0.9). The diagnostic confidence of radiologists and CNN were analyzed. There were significantly lower number of nodules rated with low confidence by CNN (19.6%) and CNN with liver metastatic status (18.1%) than two (38.4%, 44.2%, p?<?0.0001) but not a third radiologist (11.1%, p?=?0.09). We conclude that in CRC, CNN in combination with liver metastatic status equaled expert radiologists in characterizing SHHN but with better diagnostic confidence.
Project description:BACKGROUNDCurrent clinical management of patients with pulmonary nodules involves either repeated low-dose CT (LDCT)/CT scans or invasive procedures, yet causes significant patient misclassification. An accurate noninvasive test is needed to identify malignant nodules and reduce unnecessary invasive tests.METHODWe developed a diagnostic model based on targeted DNA methylation sequencing of 389 pulmonary nodule patients' plasma samples and then validation in 140 plasma samples independently. We tested the model in different stages and subtypes of pulmonary nodules.RESULTSA 100-feature model was developed and validated for pulmonary nodule diagnosis; the model achieved a receiver operating characteristic curve-AUC (ROC-AUC) of 0.843 on 140 independent validation samples, with an accuracy of 0.800. The performance was well maintained in (a) a 6 to 20 mm size subgroup (n = 100), with a sensitivity of 1.000 and adjusted negative predictive value (NPV) of 1.000 at 10% prevalence; (b) stage I malignancy (n = 90), with a sensitivity of 0.971; (c) different nodule types: solid nodules (n = 78) with a sensitivity of 1.000 and adjusted NPV of 1.000, part-solid nodules (n = 75) with a sensitivity of 0.947 and adjusted NPV of 0.983, and ground-glass nodules (n = 67) with a sensitivity of 0.964 and adjusted NPV of 0.989 at 10% prevalence. This methylation test, called PulmoSeek, outperformed PET-CT and 2 clinical prediction models (Mayo Clinic and Veterans Affairs) in discriminating malignant pulmonary nodules from benign ones.CONCLUSIONThis study suggests that the blood-based DNA methylation model may provide a better test for classifying pulmonary nodules, which could help facilitate the accurate diagnosis of early stage lung cancer from pulmonary nodule patients and guide clinical decisions.FUNDINGThe National Key Research and Development Program of China; Science and Technology Planning Project of Guangdong Province; The National Natural Science Foundation of China National.
Project description:OBJECTIVE:To develop an approach for radiology-pathology fusion of ex vivo histology of surgically excised pulmonary nodules with pre-operative CT, to radiologically map spatial extent of the invasive adenocarcinomatous component of the nodule. METHODS:Six subjects (age: 75?±?11 years) with pre-operative CT and surgically excised ground-glass nodules (size: 22.5?±?5.1 mm) with a significant invasive adenocarcinomatous component (>5 mm) were included. The pathologist outlined disease extent on digitized histology specimens; two radiologists and a pulmonary critical care physician delineated the entire nodule on CT (in-plane resolution: <0.8 mm, inter-slice distance: 1-5 mm). We introduced a novel reconstruction approach to localize histology slices in 3D relative to each other while using CT scan as spatial constraint. This enabled the spatial mapping of the extent of tumour invasion from histology onto CT. RESULTS:Good overlap of the 3D reconstructed histology and the nodule outlined on CT was observed (65.9?±?5.2%). Reduction in 3D misalignment of corresponding anatomical landmarks on histology and CT was observed (1.97?±?0.42 mm). Moreover, the CT attenuation (HU) distributions were different when comparing invasive and in situ regions. CONCLUSION:This proof-of-concept study suggests that our fusion method can enable the spatial mapping of the invasive adenocarcinomatous component from 2D histology slices onto in vivo CT. KEY POINTS:• 3D reconstructions are generated from 2D histology specimens of ground glass nodules. • The reconstruction methodology used pre-operative in vivo CT as 3D spatial constraint. • The methodology maps adenocarcinoma extent from digitized histology onto in vivo CT. • The methodology potentially facilitates the discovery of CT signature of invasive adenocarcinoma.
Project description:We test the hypothesis that a model including clinical and computed tomography (CT) features may allow discrimination between benign and malignant lung nodules in patients with soft-tissue sarcoma (STS). Seventy-one patients with STS undergoing their first lung metastasectomy were examined. The performance of multiple logistic regression models including CT features alone, clinical features alone, and combined features, was tested to evaluate the best model in discriminating malignant from benign nodules. The likelihood of malignancy increased by more than 11, 2, 6 and 7 fold, respectively, when histological synovial sarcoma sub-type was associated with the following CT nodule features: size ? 5.6 mm, well defined margins, increased size from baseline CT, and new onset at preoperative CT. Likewise, in the case of grade III primary tumor, the odds ratio (OR) increased by more than 17 times when the diameter of pulmonary nodules (PNs) was >5.6 mm, more than 13 times with well-defined margins, more than 7 times with PNs increased from baseline CT, and more than 20 times when there were new-onset nodules. Finally, when CT nodule was ?5.6 in size, it had well-defined margins, it increased in size from baseline CT, and when new onset nodules at preoperative CT were concomitant to residual primary tumor R2, the risk of malignancy increased by more than 10, 6, 25 and 28 times, respectively. The combination of clinical and CT features has the highest predictive value for detecting the malignancy of pulmonary nodules in patients with soft tissue sarcoma, allowing early detection of nodule malignancy and treatment options.