Combining radiomics and deep learning features of intra-tumoral and peri-tumoral regions for the classification of breast cancer lung metastasis and primary lung cancer with low-dose CT.

PURPOSE: To investigate the performance of deep learning and radiomics features of intra-tumoral region (ITR) and peri-tumoral region (PTR) in the diagnosing of breast cancer lung metastasis (BCLM) and primary lung cancer (PLC) with low-dose CT (LDCT). METHODS: We retrospectively collected the LDCT images of 100 breast cancer patients with lung lesions, comprising 60 cases of BCLM and 40 cases of PLC. We proposed a fusion model that combined deep learning features extracted from ResNet18-based multi-input residual convolution network with traditional radiomics features. Specifically, the fusion model adopted a multi-region strategy, incorporating the aforementioned features from both the ITR and PTR. Then, we randomly divided the dataset into training and validation sets using fivefold cross-validation approach. Comprehensive comparative experiments were performed between the proposed fusion model and other eight models, including the intra-tumoral deep learning model, the intra-tumoral radiomics model, the intra-tumoral deep-learning radiomics model, the peri-tumoral deep learning model, the peri-tumoral radiomics model, the peri-tumoral deep-learning radiomics model, the multi-region radiomics model, and the multi-region deep-learning model. RESULTS: The fusion model developed using deep-learning radiomics feature sets extracted from the ITR and PTR had the best classification performance, with the area under the curve of 0.913 (95% CI 0.840-0.960). This was significantly higher than that of the single region's radiomics model or deep learning model. CONCLUSIONS: The combination of radiomics and deep learning features was effective in discriminating BCLM and PLC. Additionally, the analysis of the PTR can mine more comprehensive tumor information.

18F-FDG PET/CT Image Deep Learning Predicts Colon Cancer Survival.

Colon cancer is a type of cancer that begins in the large intestine. In the process of efficacy evaluation, postoperative recurrence prediction and metastasis monitoring of colon cancer, traditional medical image analysis methods are highly dependent on the personal ability of the doctors. In the process of patient treatment, it not only increases the workload and work pressure for doctors, but also has some problems with traditional medical image analysis methods. Moreover, the traditional medical image analysis methods have problems such as insufficient prediction accuracy, slow prediction speed, and the risk of errors in prediction. When analyzing 18F-FDG PET/CT images by traditional medical image analysis methods, it is easy to cause problems such as untimely treatment plans and errors in diagnosis, which will adversely affect the survival of colon cancer patients. Although 18F-FDG PET/CT images have certain advantages in image clarity and accuracy compared with traditional medical imaging methods, the analysis method based on 18F-FDG PET/CT images also has certain effects in predicting the survival of colon cancer patients, but there are still many shortcomings: the 18F-FDG PET/CT image analysis method overly relies on the technical advantages of 8F-FDG PET/CT images; in the analysis and prediction of image data, it has not gotten rid of the dependence on the personal medical quality of the doctors; traditional medical image analysis methods are still used when analyzing and predicting images; there is no breakthrough in image analysis effects. In order to solve these problems, this paper combined deep learning theory, using three algorithms of the improved RBM algorithm, image feature extraction method based on deep learning, and regression neural network to analyze and predict 18F-FDG PET/CT images, and applied some algorithms to analyze and predict 18F-FDG PET/CT images, and also established a deep learning-based 18F-FDG PET/CT image survival analysis prediction model. Four aspects survival prediction accuracy, survival prediction speed, survival prediction precision, and physician satisfaction were studied through this model. The research results have shown that compared with traditional medical image analysis methods, the prediction accuracy of 18F-FDG PET/CT image survival analysis prediction model based on deep learning is improved by 0.83%, and the prediction speed is improved by 3.42%, as well as the prediction precision increased by 6.13%. The research results show that the deep learning-based 18F-FDG PET/CT image survival analysis prediction model established in this paper is of great significance to improve the survival rate of colon cancer patients, and also promotes the development of the medical industry.

Prognostic significance of volume-based 18F-FDG PET/CT parameters and correlation with PD-L1 expression in patients with surgically resected lung adenocarcinoma.

ABSTRACT: The aim of this study was to retrospectively analyze 18F-FDG positron emission tomography/computed tomography (18F-FDG PET/CT) metabolic variables, programmed death-ligand 1 (PD-L1) and phosphorylated signal transducer and activator of transcription 3 (p-STAT3) tumor expression, and other factors as predictors of disease-free survival (DFS) in patients with lung adenocarcinoma (LUAD) (stage IA-IIIA) who underwent surgical resection. We still lack predictor of immune checkpoint (programmed cell death-1 [PD-1]/PD-L1) inhibitors. Herein, we investigated the correlation between metabolic parameters from 18F-FDG PET/CT and PD-L1 expression in patients with surgically resected LUAD.Seventy-four patients who underwent 18F-FDG PET/CT prior to treatment were consecutively enrolled. The main 18F-FDG PET/CT-derived variables were primary tumor maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG). Surgical tumor specimens were analyzed for PD-L1 and p-STAT3 expression using immunohistochemistry. Correlations between immunohistochemistry results and 18F-FDG PET/CT-derived variables were compared. Associations of PD-L1 and p-STAT3 tumor expression, 18F-FDG PET/CT-derived variables, and other factors with DFS in resected LUAD were evaluated.All tumors were FDG-avid. The cutoff values of low and high SUVmax, MTV, and TLG were 12.60, 14.87, and 90.85, respectively. The results indicated that TNM stage, PD-L1 positivity, and high 18F-FDG PET/CT metabolic volume parameters (TLG ≥90.85 or MTV ≥14.87) were independent predictors of worse DFS in resected LUAD. No 18F-FDG metabolic parameters associated with PD-L1 expression were observed (chi-square test), but we found that patients with positive PD-L1 expression have significantly higher SUVmax (P = .01), MTV (P = .00), and TLG (P = .00) than patients with negative PD-L1 expression.18F-FDG PET/CT metabolic volume parameters (TLG ≥90.85 or MTV ≥14.87) were more helpful in prognostication than the conventional parameter (SUVmax), PD-L1 expression was an independent predictor of DFS in patients with resected LUAD. Metabolic parameters on 18F-FDG PET/CT have a potential role for 18F-FDG PET/CT in selecting candidate LUAD for treatment with checkpoint inhibitors.

Comparison of machine learning methods for classifying mediastinal lymph node metastasis of non-small cell lung cancer from 18F-FDG PET/CT images.

BACKGROUND: This study aimed to compare one state-of-the-art deep learning method and four classical machine learning methods for classifying mediastinal lymph node metastasis of non-small cell lung cancer (NSCLC) from 18F-FDG PET/CT images. Another objective was to compare the discriminative power of the recently popular PET/CT texture features with the widely used diagnostic features such as tumor size, CT value, SUV, image contrast, and intensity standard deviation. The four classical machine learning methods included random forests, support vector machines, adaptive boosting, and artificial neural network. The deep learning method was the convolutional neural networks (CNN). The five methods were evaluated using 1397 lymph nodes collected from PET/CT images of 168 patients, with corresponding pathology analysis results as gold standard. The comparison was conducted using 10 times 10-fold cross-validation based on the criterion of sensitivity, specificity, accuracy (ACC), and area under the ROC curve (AUC). For each classical method, different input features were compared to select the optimal feature set. Based on the optimal feature set, the classical methods were compared with CNN, as well as with human doctors from our institute. RESULTS: For the classical methods, the diagnostic features resulted in 81~85% ACC and 0.87~0.92 AUC, which were significantly higher than the results of texture features. CNN's sensitivity, specificity, ACC, and AUC were 84, 88, 86, and 0.91, respectively. There was no significant difference between the results of CNN and the best classical method. The sensitivity, specificity, and ACC of human doctors were 73, 90, and 82, respectively. All the five machine learning methods had higher sensitivities but lower specificities than human doctors. CONCLUSIONS: The present study shows that the performance of CNN is not significantly different from the best classical methods and human doctors for classifying mediastinal lymph node metastasis of NSCLC from PET/CT images. Because CNN does not need tumor segmentation or feature calculation, it is more convenient and more objective than the classical methods. However, CNN does not make use of the import diagnostic features, which have been proved more discriminative than the texture features for classifying small-sized lymph nodes. Therefore, incorporating the diagnostic features into CNN is a promising direction for future research.

[Correlation between metabolic tumor volume (MTV) and microvessel density (MVD) and blood-borne metastasis in colorectal carcinoma].

OBJECTIVE: To explore the correlation between metabolic tumour volume (MTV) and microvessel density (MVD) and blood-borne metastasis in colorectal carcinoma. METHODS: Thirty-six patients with CRC conformed by pathology underwent PET-CT examination before operation. SUVmax and MTV were obtained by PET VCRA software. The blood vessels were identified with CD34 immunohistochemical staining, and the MVD was recorded. The correlation between SUVmax and MTV with histological differentiation, T stage, MVD and blood-borne metastasis was analyzed. RESULTS: The SUVmax, MTV and MVD in patients with blood-borne metastasis were 5.15 ± 5.41, (22.99 ± 18.63) cm³ and 14.17 ± 3.63, and were 10.65 ± 3.79, (16.95 ± 11.82) cm³ and 11.27 ± 3.69, respectively, in patients with non-blood-borne metastasis. The differences of SUVmax, MTV and MVD between blood-borne metastasis and non-blood-borne metastasis patients were statistically significant (all P > 0.05). Pearson correlation analysis found that there was no linear correlation between SUVmax and MVD, and the SUVmax was not statistically significant between high and low MVD groups (t = 0.919, P = 0.364). But there was a linear correlation between MTV and MVD (r = 0.621, P = 0.000), and the MTV was statistically significant between high and low MVD groups (t = 3.567, P = 0.001). The receiver-operating characteristic curves showed that MTV could be used to predict blood-borne metastasis of CRC, and the best cutoff value for MTV was 14.975 cm³, and the sensitivity, specificity, negative predictive value and positive predictive value were 85.7%, 54.5%, 72.3% and 64.2%, respectively. There were no significant relationships between SUVmax, MTV, MVD, blood-borne metastasis and histological differentiation (P > 0.05). With the increased T stage, the MTV, MVD and the probability of blood-borne metastasis were also increased (all P < 0.05). CONCLUSIONS: There are correlations between MTV and MVD and blood-borne metastasis in CRC. The risk of blood-borne metastasis in patients with MTV > 14.975 cm³ is higher, and needs to take more effective intervention.

[Density and SUV ratios from PET/CT in the detection of mediastinal lymph node metastasis in non-small cell lung cancer].

BACKGROUND AND OBJECTIVE: Mediastinal involvement in lung cancer is a highly significant prognostic factor for survival, and accurate staging of the mediastinum will correctly identify patients who will benefit the most from surgery. Positron emission tomography/computed tomography (PET/CT) has become the standard imaging modality for the staging of patients with lung cancer. The aim of this study is to investigate 18-fluoro-2-deoxy-glucose (18F-FDG) PET/CT imaging in the detection of mediastinal disease in lung cancer. METHODS: A total of 72 patients newly diagnosed with non-small cell lung cancer (NSCLC) who underwent preoperative whole-body 18F-FDG PET/CT were retrospectively included. All patients underwent radical surgery and mediastinal lymph node dissection. Mediastinal disease was histologically confirmed in 45 of 413 lymph nodes. PET/CT doctors analyzed patients' visual images and evaluated lymph node's short axis, lymph node's maximum standardized uptake value (SUVmax), node/aorta density ratio, node/aorta SUV ratio, and other parameters using the histopathological results as the reference standard. The optimal cutoff value for each ratio was determined by receiver operator characteristic curve analysis. RESULTS: Using a threshold of 0.9 for density ratio and 1.2 for SUV ratio yielded high accuracy for the detection of mediastinal disease. The lymph node's short axis, lymph node's SUVmax, density ratio, and SUV ratio of integrated PET/CT for the accuracy of diagnosing mediastinal lymph node was 95.2%. The diagnostic accuracy of mediastinal lymph node with conventional PET/CT was 89.8%, whereas that of PET/CT comprehensive analysis was 90.8%. CONCLUSIONS: Node/aorta density ratio and SUV ratio may be complimentary to conventional visual interpretation and SUVmax measurement. The use of lymph node's short axis, lymph node's SUVmax, and both ratios in combination is better than either conventional PET/CT analysis or PET/CT comprehensive analysis in the assessment of mediastinal disease in NSCLC patients.
.

Use of 18F-FDG PET Scan and Ultrasound-guided Biopsy in the Diagnosis of Hepatic Carcinomas.

BACKGROUND/AIMS: PET is a highly specific tool in liver cancer diagnosis. However, in small hepatic lesions, typical imaging characteristics are lacking. The aim of this study was to investigate the accuracy of PET-CT and ultrasound-guided biopsy for diagnosis of liver carcinomas. METHODOLOGY: Clinical data of 58 patients who were diagnosed and treated in our hospital from January 2008 to December 2012 were retrospectively reviewed. Before surgery, patients received PET-CT assessment, ultrasound examination, and ultrasound-guided biopsy of liver tissues. Forty three patients with cancer underwent partial hepatectomy. Clinical data were collected from hospital records and follow-up questionnaires. RESULTS: Postoperative histology revealed hepatocellular carcinoma in 26 patients (46.4%), liver metastases in 12 patients (21.4%). Only 4 patients (7.1%) had cholangiocarcinoma. The sensitivity of PET-CT, ultrasound, and ultrasound-guided biopsy for diagnosis of hepatocellular carcinoma (HCC) was 92.9%, 95.5%, and 100%, respectively. The surgery-related complication rate was 3.6%. Prognosis was good, with 1- and 3-year survival rates of 83.4% and 67.9%, respectively. CONCLUSIONS: PET-CT is highly specific for diagnosis of hepatic cancer, which is consistent with the diagnosis of ultrasound-guided biopsy and postoperative histological assessments. Combination use of PET-CT, ultrasound assessment and ultrasound-guided biopsy will provide the most accurate diagnosis of liver cancer.

The method and efficacy of support vector machine classifiers based on texture features and multi-resolution histogram from (18)F-FDG PET-CT images for the evaluation of mediastinal lymph nodes in patients with lung cancer.

OBJECTIVES: In clinical practice, image analysis is dependent on simply visual perception and the diagnostic efficacy of this analysis pattern is limited for mediastinal lymph nodes in patients with lung cancer. In order to improve diagnostic efficacy, we developed a new computer-based algorithm and tested its diagnostic efficacy. METHODS: 132 consecutive patients with lung cancer underwent (18)F-FDG PET/CT examination before treatment. After all data were imported into the database of an on-line medical image analysis platform, the diagnostic efficacy of visual analysis was first evaluated without knowing pathological results, and the maximum short diameter and maximum standardized uptake value (SUVmax) were measured. Then lymph nodes were segmented manually. Three classifiers based on support vector machine (SVM) were constructed from CT, PET, and combined PET-CT images, respectively. The diagnostic efficacy of SVM classifiers was obtained and evaluated. RESULTS: According to ROC curves, the areas under curves for maximum short diameter and SUVmax were 0.684 and 0.652, respectively. The areas under the ROC curve for SVM1, SVM2, and SVM3 were 0.689, 0.579, and 0.685, respectively. CONCLUSION: The algorithm based on SVM was potential in the diagnosis of mediastinal lymph nodes.

Evaluation of the diagnostic performance of magnetic resonance spectroscopy in brain tumors: a systematic review and meta-analysis.

OBJECT: The aim of this study was to determine the suitability of magnetic resonance spectroscopy (MRS) for screening brain tumors, based on a systematic review and meta-analysis of published data on the diagnostic performance of MRS. METHODS: The PubMed and PHMC databases were systematically searched for relevant studies up to December 2013. The sensitivities and specificities of MRS in individual studies were calculated and the pooled diagnostic accuracies, with 95% confidence intervals (CI), were assessed under a fixed-effects model. RESULTS: Twenty-four studies were included, comprising a total of 1013 participants. Overall, no heterogeneity of diagnostic effects was observed between studies. The pooled sensitivity and specificity of MRS were 80.05% (95% CI = 75.97%-83.59%) and 78.46% (95% CI: 73.40%-82.78%), respectively. The area under the summary receiver operating characteristic curve was 0.78. Stratified meta analysis showed higher sensitivity and specificity in child than adult. CSI had higher sensitivity and SV had higher specificity. Higher sensitivity and specificity were obtained in short TE value. CONCLUSION: Although the qualities of the studies included in the meta-analysis were moderate, current evidence suggests that MRS may be a valuable adjunct to magnetic resonance imaging for diagnosing brain tumors, but requires selection of suitable technique and TE value.

A correlation study between maximum standardized uptake values and pathology and clinical staging in nonsmall cell lung cancer.

OBJECTIVE: To investigate the relationship between maximum standardized uptake value and pathological type, degree of differentiation, tumor size, and clinical staging of nonsmall cell lung cancer (NSCLC). METHODS: This study included 135 cases with pathologically proven NSCLC. Correlations between maximum standardized uptake value (SUVmax) and pathological type, degree of differentiation, tumor size, and clinical staging were analyzed. RESULTS: There was a significant correlation between the SUVmax of NSCLC and the pathological type (r= 0.391, P= 0.000); the SUVmax of squamous cell carcinoma (SCC) was higher than that of adenocarcinoma (AC) (P =0.000), and the SUVmax of AC was higher than that of bronchioloalveolar carcinoma (P = 0.004). There was a positive correlation between the SUVmax of AC and the degree of differentiation (r= 0.222, P = 0.044); SUVmax was lower in well-differentiated ACs than in moderately or poorly differentiated ACs (P=0.034 and 0.022 respectively); however, there was no statistical difference between the moderately differentiated and poorly differentiated groups (P= 1.000). There was no correlation between the SUVmax of SCC and the degree of differentiation (r= - 0.304, P= 0.054). A positive correlation was found between the SUVmax of NSCLC and tumor size (r= 0.569, P =0.000). The SUVmax of AC had a positive correlation with clinical staging (r= 0.298, P = 0.006); SUVmax was lower in stage I than in stages II, III, and IV (P = 0.047, 0.038 and 0.015, respectively); however, the SUVmax in stages II, III, and IV were not different (P= 0.708, 0.570 and 0.528, respectively). There was no correlation between the SUVmax of SCC and clinical staging (r =0.066, P = 0.680). CONCLUSION: There was a correlation between the SUVmax of NSCLC and the pathological type and tumor size. A positive correlation was found between the SUVmax of AC and the degree of differentiation and clinical staging. There were no correlations between the SUVmax of SCC and the degree of differentiation or clinical staging.