Rectal Cancer Stages T2 and T3 Identification Based on Asymptotic Hybrid Feature Maps

Many existing intelligent recognition technologies require huge datasets for model learning.However,it is not easy to collect rectal cancer images,so the performance is usually low with limited training samples.In addition,traditional rectal cancer staging is time-consuming,error-prone,and susceptible to physicians’subjective awareness as well as professional expertise.To settle these deficiencies,we propose a novel deep-learning model to classify the rectal cancer stages of T2 and T3.First,a novel deep learning model(RectalNet)is constructed based on residual learning,which combines the squeeze-excitation with the asymptotic output layer and new cross-convolution layer links in the residual block group.Furthermore,a two-stage data augmentation is designed to increase the number of images and reduce deep learning’s dependence on the volume of data.The experiment results demonstrate that the proposed method is superior to many existing ones,with an overall accuracy of 0.8583.Oppositely,other traditional techniques,such as VGG16,DenseNet121,EL,and DERNet,have an average accuracy of 0.6981,0.7032,0.7500,and 0.7685,respectively.

FLAIR in Gd-EOB-DTPA Enhanced MRI: Does It Make a Difference?

Background: Gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA;gadoxetic acid disodium, Primovist, Bayer Healthcare, Berlin, Germany) is a gadolinium based contrast agent with hepatocyte specific properties. In patients scanned for hepatic metastasis using Gd-EOB-DTPA, it is important to differentiate hepatic metastasis with cysts and hemangiomas, which are the two most common benign lesions seen in the liver. Yet, in some cases it is difficult to differentiate these lesions. Purpose: The purpose of this study was to retrospectively investigate the usefulness of combining Fluid-attenuated inversion recovery (FLAIR) with Gd-EOB-DTPA enhanced MRI. Material and Methods: Gd-EOB-DTPA enhanced MRIs of 47 patients (19 male, 27 female) with a mean age of 68 years (range 32 - 85 years old) with a total of 121 lesions (68 cysts, 37 metastasis, 16 hemangiomas) were included in the study. T1WI, T2WI, heavy T2WI, dynamic contrast enhanced MRI, and FLAIR images of these lesions were evaluated. The patients were randomly divided into two groups (Groups A and B), and two independent radiologists were asked to give a diagnosis for each lesion. The radiologists were allowed to view FLAIR images for only Group B. Diagnostic performance regarding the differentiation of cysts, hemangiomas and metastases was assessed. MRI examinations were scanned using a 1.5 Tesla system (Echlon Vega, Hitachi,) with an 8 channel multiple array coil (RAPID body coil). Results: An statistically significant improvement (p < 0.05) of the specificity for cysts was seen from 71.9% (Group A) to 90.9% (Group B) for Reader 1, and 75.0% (Group A) to 93.3% (Group B) for Reader 2. No statistical differences were seen between the two groups for sensitivity and specificity of hemangiomas. Although no statistical difference was seen between the two groups, an improvement (77.8 in Group A to 97.2 in Group B for Reader 1, and 85.7 in Group A to 100 in Group B for Reader 2) was seen for the sensitivity of metastasis with the addition of FLAIR. Conclusion: An improvement of diagnostic accuracy, especially for cysts, was seen with the addition of FLAIR to Gd-EOB-DTPA enhanced MRI.

Prostate cancer identification: quantitative analysis of T2-weighted MR images based on a back propagation artificial neural network model

Computer-aided diagnosis(CAD) systems have been proposed to assist radiologists in making diagnostic decisions by providing helpful information. As one of the most important sequences in prostate magnetic resonance imaging(MRI), image features from T2-weighted images(T2WI) were extracted and evaluated for the diagnostic performances by using CAD. We extracted 12 quantitative image features from prostate T2-weighted MR images. The importance of each feature in cancer identification was compared in the peripheral zone(PZ) and central gland(CG), respectively. The performance of the computer-aided diagnosis system supported by an artificial neural network was tested. With computer-aided analysis of T2-weighted images, many characteristic features with different diagnostic capabilities can be extracted. We discovered most of the features(10/12) had significant difference(P<0.01) between PCa and non-PCa in the PZ, while only five features(sum average, minimum value, standard deviation, 10 th percentile, and entropy) had significant difference in CG. CAD prediction by features from T2 w images can reach high accuracy and specificity while maintaining acceptable sensitivity. The outcome is convictive and helpful in medical diagnosis.