Application of convolutional neural network in the diagnosis of the invasion depth of gastric cancer based on conventional endoscopy.

BACKGROUND AND AIMS: According to guidelines, endoscopic resection should only be performed for patients whose early gastric cancer invasion depth is within the mucosa or submucosa of the stomach regardless of lymph node involvement. The accurate prediction of invasion depth based on endoscopic images is crucial for screening patients for endoscopic resection. We constructed a convolutional neural network computer-aided detection (CNN-CAD) system based on endoscopic images to determine invasion depth and screen patients for endoscopic resection. METHODS: Endoscopic images of gastric cancer tumors were obtained from the Endoscopy Center of Zhongshan Hospital. An artificial intelligence-based CNN-CAD system was developed through transfer learning leveraging a state-of-the-art pretrained CNN architecture, ResNet50. A total of 790 images served as a development dataset and another 203 images as a test dataset. We used the CNN-CAD system to determine the invasion depth of gastric cancer and evaluated the system's classification accuracy by calculating its sensitivity, specificity, and area under the receiver operating characteristic curve. RESULTS: The area under the receiver operating characteristic curve for the CNN-CAD system was .94 (95% confidence interval [CI], .90-.97). At a threshold value of .5, sensitivity was 76.47%, and specificity 95.56%. Overall accuracy was 89.16%. Positive and negative predictive values were 89.66% and 88.97%, respectively. The CNN-CAD system achieved significantly higher accuracy (by 17.25%; 95% CI, 11.63-22.59) and specificity (by 32.21%; 95% CI, 26.78-37.44) than human endoscopists. CONCLUSIONS: We constructed a CNN-CAD system to determine the invasion depth of gastric cancer with high accuracy and specificity. This system distinguished early gastric cancer from deeper submucosal invasion and minimized overestimation of invasion depth, which could reduce unnecessary gastrectomy.

Early Evaluation of Relative Changes in Tumor Stiffness by Shear Wave Elastography Predicts the Response to Neoadjuvant Chemotherapy in Patients With Breast Cancer.

OBJECTIVES: Neoadjuvant chemotherapy plays an important role in comprehensive therapy for breast cancer, but response prediction is imperfect. Shear wave elastography (SWE) is a novel technique that can quantitatively evaluate tissue stiffness. In this study, we sought to investigate the application value of SWE for early prediction of the response to neoadjuvant chemotherapy in patients with breast cancer. METHODS: We prospectively evaluated tumor stiffness in 62 patients with breast cancer using SWE, which was performed at baseline and after the second cycle of neoadjuvant chemotherapy. After chemotherapy, all of the patients underwent surgery. We investigated the correlations between the relative changes in tumor stiffness (Δ stiffness) after 2 cycles of chemotherapy and the pathologic response to the therapy. RESULTS: Compared with baseline values, tumor stiffness after 2 cycles of neoadjuvant chemotherapy was significantly decreased in responders (P < .001) but not in nonresponders (P = .172). The Δstiffness was significantly higher in responders (-42.194%) than in nonresponders (-23.593%; P = .001). As determined at either the baseline or after the second cycle of chemotherapy, tumor stiffness was significantly lower in responders than in nonresponders (P = .033 and .009, respectively). The Δ stiffness threshold for distinguishing between responders and nonresponders was -36.1% (72.92% sensitivity and 85.71% specificity). Furthermore, correlating Δ stiffness with clinical and pathologic characteristics, we found that estrogen and progesterone receptor expression showed statistically significant correlations with Δ stiffness (estrogen receptor, P = .008; progesterone receptor, P = .023). CONCLUSIONS: Early evaluation of relative changes in tumor stiffness using SWE could effectively predict the response to neoadjuvant chemotherapy in patients with breast cancer and might indicate better therapeutic strategies on a timelier basis.

Shear wave elastography combined with the thyroid imaging reporting and data system for malignancy risk stratification in thyroid nodules.

To retrospectively evaluate the diagnostic performance of shear wave elastography (SWE) and thyroid imaging reporting and data system (TI-RADS) in differentiating malignant and benign thyroid nodules. A total of 313 thyroid nodules in 227 patients were included. All thyroid nodules were underwent SWE and TI-RADS before fine needle aspiration biopsy and/or surgery. SWE elasticity indices of the maximum (Emax), mean (Emean), minimum (Emin) and elastic ratio (ER) in thyroid nodules were measured. Nodules with solid component, marked hypoechogenicity, poorly defined margins, micro-calcifications, and a taller-than-wide shape were classified as suspicious at gray-scale ultrasonography. The level of TI-RADS was determined according to the number of suspicious ultrasonography features. The combined methods of SWE and TI-RADS in thyroid nodules were calculated. In the 313 nodules, 194 were malignant, and 119 were benign. SWE and TI-RADS were significantly higher in malignant nodules than benign nodules (P < 0.001). The most accurate SWE cut-off value, 51.95 kPa for Emax, achieved a sensitivity of 81.44% and a specificity of 83.19% for discriminating malignant nodules from benign nodules. There are two methods in combination with SWE and TI-RADS. The one is "tandem" method, which has a higher specificity (95.80%), positive likelihood ratio (18.16) and positive predictive value (96.73%). The other one is "parallel" method, which shows sensitivity (94.85%), negative likelihood ratio (0.07) and negative predictive value (90.00%).We believe that the methods could be used as a simple tool to stratify the risk of thyroid nodules accurately.

In vivo experimental study on the effects of fluid in increasing the efficiency of radiofrequency ablation.

BACKGROUND: Radiofrequency ablation (RFA) is the most widely used and studied method internationally for the local treatment of liver tumors. However, the extension of coagulation necrosis in one RFA procedure is limited and incomplete coverage of the damaged area can lead to a high local recurrence rate. OBJECTIVE: In this study, we compared the effects of different solutions in enhancing hepatic radiofrequency by establishing a rabbit VX2 liver cancer model. We also determined the optimal solution to maximise effects on the extent of RFA-induced coagulation necrosis. METHODS: Thirty VX2 tumor rabbits were randomly assigned to five groups: group A, RFA alone; group B, RFA with anhydrous ethanol injection; group C, RFA with 5% hypertonic saline injection; group D, RFA with lidocaine injection; and group E, RFA with a mixed solution. Routine ultrasound examinations and contrast-enhanced ultrasound (CEUS) of the ablation areas were performed after RFA. Then, we measured the major axis and transverse diameter and compared the areas of coagulation necrosis induced by RFA. RESULTS: The mean ablation area range increased in groups B, C and especially E, and the scopes were greater compared with group A. Preoperative application of anhydrous ethanol, hypertonic saline, lidocaine and the mixed solution (groups B, C, D and E, respectively) resulted in larger coagulation necrosis areas than in group A (p<0.05). Among the groups, the coagulation necrosis areas in group E was largest, and the difference was statistically significant compared with other groups (p<0.05). Pathological findings were consistent with imaging results. CONCLUSIONS: A mixture of dehydrated alcohol, hypertonic saline and lidocaine injected with RFA increases the extent of coagulation necrosis in the liver with a single application, and the mixed solution is more effective than any other injection alone.

Shorter distance between the nodule and capsule has greater risk of cervical lymph node metastasis in papillary thyroid carcinoma.

BACKGROUND: The purpose of this study was to assess the relationship between different sonographic features of papillary thyroid carcinoma (PTC) on high-frequency ultrasound and cervical lymph node metastasis (CLNM). MATERIALS AND METHODS: We enrolled 548 patients who underwent initial surgery for PTC between May 2011 and December 2012 in our hospital at diagnosis. The sonographic features of 513 PTC nodules in 513 eligible patients, who had single PTC nodules in their thyroid glands, were retrospectively investigated. All patients with a suspect malignant nodule (d<0.5cm) among multiple nodules were initially diagnosed by fine-needle aspiration biopsy (FNAB) to ascertain if the suspect nodule was PTC. The final diagnosis of all the thyroid nodules and existence of CLNM were based on postoperative pathology. Patients were divided into two groups: a positive group with CLNM (224 nodules) and a negative group without CLNM (289 nodules). The following factors were investigated: gender, age, echogenicity, echotexture, size, shape, location, margin, contour, calcification morphology, distance between the nodule and pre- or post-border of the thyroid capsule, vascularity and the differences between the two groups. RESULTS: Correlation analysis showed that shorter distances between the nodule and pre- or post- border of thyroid capsule resulted in greater risk of CLNM (Spearman correlation coefficient=-0.22, p<0.0001). The significant factors in multivariate analysis were age<45yrs, larger size (d>1cm), "wider than tall" shape, extrathyroid extension and mixed flow (internal and peripheral) (p<0.05, OR=0.406, 2.093, 0.461, 1.610, 1.322). CONCLUSIONS: Significant sonographic features of PTC nodules in preoperative high-frequency ultrasound are crucial for predicting CLNM.

Expression of Fas/FasL in CD8+ T and CD3+ Foxp3+ Treg cells--relationship with apoptosis of circulating CD8+ T cells in hepatocellular carcinoma patients.

AIMS: Dysfunction of the host immune system in cancer patients can be due to a number of factors, including lymphocyte apoptosis. Several studies showed that Foxp3+T cells take part in inducing this process by expressing FasL in tumor patients. However, the relationship between apoptosis, CD8+T cells and Foxp3+T cells in HCC patients is still unclear. The present study was designed to investigate the correlation between apoptosis levels and Fas/FasL expression in CD8+T lymphocytes and Foxp3+T cells in patients with HCC. METHODS: CD8+T cells and CD3+Foxp3+T cells were tested from peripheral blood of HCC patients and normal controls and subjected to multicolor flow cytometry. The expression of an apoptosis marker (annexin V) and the death receptor Fas in CD8+T cells and FasL in CD3+Foxp3+T cells were evaluated. Serum TGF-ß1 levels in patients with HCC were measured by enzyme-linked immunosorbent assay. The relationship between apoptosis and Fas expression, as well as FasL expression in CD3+Foxp3+T cells was then evaluated. RESULTS: The frequency of CD8+T cells binding annexin V and Fas expression in CD8+T cells, were all higher in HCC patients than normal controls and the proportion of apoptotic CD8+T cells correlated with their Fas expression. Serum TGF-ß1 levels correlated inversely with CD3+Foxp3+T cells. CONCLUSIONS: Fas/FasL interactions might lead to excessive turnover of CD8+T cells and reduce anti-tumor immune responses in patients with HCC. Further investigations of apoptosis induction in Fas+CD8+T cells in vitro are required.