Fast and Accurate U-Net Model for Fetal Ultrasound Image Segmentation

U-Net based algorithms, due to their complex computations, include limitations when they are used in clinical devices. In this paper, we addressed this problem through a novel U-Net based architecture that called fast and accurate U-Net for medical image segmentation task. The proposed fast and accurate U-Net model contains four tuned 2D-convolutional, 2D-transposed convolutional, and batch normalization layers as its main layers. There are four blocks in the encoder-decoder path. The results of our proposed architecture were evaluated using a prepared dataset for head circumference and abdominal circumference segmentation tasks, and a public dataset (HC18-Grand challenge dataset) for fetal head circumference measurement. The proposed fast network significantly improved the processing time in comparison with U-Net, dilated U-Net, R2U-Net, attention U-Net, and MFP U-Net. It took 0.47 seconds for segmenting a fetal abdominal image. In addition, over the prepared dataset using the proposed accurate model, Dice and Jaccard coefficients were 97.62% and 95.43% for fetal head segmentation, 95.07%, and 91.99% for fetal abdominal segmentation. Moreover, we have obtained the Dice and Jaccard coefficients of 97.45% and 95.00% using the public HC18-Grand challenge dataset. Based on the obtained results, we have concluded that a fine-tuned and a simple well-structured model used in clinical devices can outperform complex models.

A Generative Adversarial Network Fused with Dual-Attention Mechanism and Its Application in Multitarget Image Fine Segmentation

Aiming at the problem of insignificant target morphological features, inaccurate detection and unclear boundary of small-target regions, and multitarget boundary overlap in multitarget complex image segmentation, combining the image segmentation mechanism of generative adversarial network with the feature enhancement method of nonlocal attention, a generative adversarial network fused with attention mechanism (AM-GAN) is proposed. The generative network in the model is composed of residual network and nonlocal attention module, which use the feature extraction and multiscale fusion mechanism of residual network, as well as feature enhancement and global information fusion ability of nonlocal spatial-channel dual attention to enhance the target features in the detection area and improve the continuity and clarity of the segmentation boundary. The adversarial network is composed of fully convolutional networks, which penalizes the loss of information in small-target regions by judging the authenticity of prediction and label segmentation and improves the detection ability of the generative adversarial model for small targets and the accuracy of multitarget segmentation. AM-GAN can use the GAN’s inherent mechanism that reconstruct and repair high-resolution image, as well as the ability of nonlocal attention global receptive field to strengthen detail features, automatically learn to focus on target structures of different shapes and sizes, highlight salient features useful for specific tasks, reduce the loss of image detail features, improve the accuracy of small-target detection, and optimize the segmentation boundary of multitargets. Taking medical MRI abdominal image segmentation as a verification experiment, multitargets such as liver, left/right kidney, and spleen are selected for segmentation and abnormal tissue detection. In the case of small and unbalanced sample datasets, the class pixels’ accuracy reaches 87.37%, the intersection over union is 92.42%, and the average Dice coefficient is 93%. Compared with other methods in the experiment, the segmentation precision and accuracy are greatly improved. It shows that the proposed method has good applicability for solving typical multitarget image segmentation problems such as small-target feature detection, boundary overlap, and offset deformation.

Evaluation of Convolutional Neural Network Architecture for Feasibility Analysis on Fetal Abdomen and Brain Images

This research aims to evaluate the possibilities of fetus ultrasound image classification using machine learning algorithms as normal or abnormal. Most of the earlier research works have produced a high percentage of false-negative classification results—recent research work aimed to reduce the rate of false-negative diagnoses. Also, the number of sonologists for analyzing prenatal ultrasound worldwide is very less and solved by developing an efficient algorithm, which reduces the percentage of false negatives in the diagnosis output. Several earlier research works focused on analyzing fetal abdominal image or fetal head images, making the medical industry use two different diagnostic modules separately. This work aims to design and implement a convolution frame-work named as two Convolution Neural Network (tCNN) model for diagnosing any fetal images. The proposed tCNN model diagnoses the fetal abdominal and fetal brain images and classify them as normal or abnormal. CNN1 of tCNN performs segmentation and classification based on the acceptance of abdomen circumference and stomach bubble, umbilical vein, and amniotic fluid measurements. CNN2 shows based on head circumference and head and abdominal circumference, femur, crown-rump, and humerus lengths measured.With clinical validation, an extensive experiment carried out and the results compared with the experts in terms of segmentation accuracy and the obstetric measurements. This paper provides a foundation for future multi-classification research works on diagnosing fetal intracranial abnormalities and differential diagnosis using machine learning algorithms.

GMP-compliant production of [68Ga]Ga-NeoB for positron emission tomography imaging of patients with gastrointestinal stromal tumor

Background [68Ga]Ga-NeoB is a novel DOTA-coupled Gastrin Releasing Peptide Receptor (GRPR) antagonist with high affinity for GRPR and good in vivo stability. This study aimed at (1) the translation of preclinical results to the clinics and establish the preparation of [68Ga]Ga-NeoB using a GMP conform kit approach and a licensed 68Ge/68Ga generator and (2) to explore the application of [68Ga]Ga-NeoB in patients with gastrointestinal stromal tumors (GIST) before and/or after interventional treatment (selective internal radiotherapy, irreversible electroporation, microwave ablation). Results Validation of the production and quality control of [68Ga]Ga-NeoB for patient use had to be performed before starting the GMP production. Six independent batches of [68Ga]Ga-NeoB were produced, all met the quality and sterility criteria and yielded 712 ± 73 MBq of the radiotracer in a radiochemical purity of > 95% and a molar activity of 14.2 ± 1.5 GBq/μmol within 20 min synthesis time and additional 20 min quality control. Three patients (2 females, 1 male, 51–77 yrs. of age) with progressive gastrointestinal stromal tumor metastases in the liver or peritoneum not responsive to standard tyrosine kinase inhibitor therapy underwent both [68Ga]Ga-NeoB scans prior and after interventional therapy. Radiosynthesis of 68Ga-NeoB was performed using a kit approach under GMP conditions. No specific patient preparation such as fasting or hydration was required for [68Ga]Ga-NeoB PET/CT imaging. Contrast-enhanced PET/CT studies were performed. A delayed, second abdominal image after the administration of the of [68Ga]Ga-NeoB was acquired at 120 min post injection. Conclusions A fully GMP compliant kit preparation of [68Ga]Ga-NeoB enabling the routine production of the tracer under GMP conditions was established for clinical routine PET/CT imaging of patients with metastatic GIST and proved to adequately visualize tumor deposits in the abdomen expressing GRPR. Patients could benefit from additional information derived from [68Ga]Ga-NeoB diagnosis to assess the presence of GRPR in the tumor tissue and monitor antitumor treatment.

Application of Abdominal Imaging Based on Nano Drug Delivery System for Diagnosis and Treatment of Liver Cancer

Liver cancer has a high incidence and a poor prognosis, which seriously affects human health. Doxorubicin is one of the chemotherapeutics used in the treatment of tumours, but its severe adverse reactions, especially cardiac toxicity, have limited its clinical application. The nanometre drug delivery system enables drug-loaded nanoparticles to be specifically concentrated in tumour tissues, increasing cell uptake and improving curative effect. Therefore, in this paper, folic acid-modified mesoporous silica nanoparticles (MSN-NH2-PEG-FA) were synthesized by modifying the folic acid on the surface of a drug carrier by using the characteristics of the expression of folic acid receptors, and using it as a drug. The carrier was loaded with antitumor drug doxorubicin hydrochloride (DOX), and a nanometre drug delivery system (MSN-NH2-PEG-FA/DOX) was constructed. At the same time, the near-infrared dye Cy5 was used to mark the mother nucleus to construct fluorescent nanoparticles (MSN-NH2-PEG-FA/DOX-Cy5) for cell and tumour imaging, so as to obtain the abdominal image of liver cancer patients, thereby realizing diagnosis and treatment. The research results show that the carrier can specifically gather in the liver area, reduce the distribution in the heart, reduce the toxic and side effects of drugs, and prolong the survival time of patients. The results of this study provide new ideas for the treatment of liver cancer, and provide a new theoretical basis and experimental basis for the study of inorganic nanomaterials as targeted drug delivery systems.

Generative multi-adversarial network for striking the right balance in abdominal image segmentation

Purpose The identification of abnormalities that are relatively rare within otherwise normal anatomy is a major challenge for deep learning in the semantic segmentation of medical images. The small number of samples of the minority classes in the training data makes the learning of optimal classification challenging, while the more frequently occurring samples of the majority class hamper the generalization of the classification boundary between infrequently occurring target objects and classes. In this paper, we developed a novel generative multi-adversarial network, called Ensemble-GAN, for mitigating this class imbalance problem in the semantic segmentation of abdominal images. Method The Ensemble-GAN framework is composed of a single-generator and a multi-discriminator variant for handling the class imbalance problem to provide a better generalization than existing approaches. The ensemble model aggregates the estimates of multiple models by training from different initializations and losses from various subsets of the training data. The single generator network analyzes the input image as a condition to predict a corresponding semantic segmentation image by use of feedback from the ensemble of discriminator networks. To evaluate the framework, we trained our framework on two public datasets, with different imbalance ratios and imaging modalities: the Chaos 2019 and the LiTS 2017. Result In terms of the F1 score, the accuracies of the semantic segmentation of healthy spleen, liver, and left and right kidneys were 0.93, 0.96, 0.90 and 0.94, respectively. The overall F1 scores for simultaneous segmentation of the lesions and liver were 0.83 and 0.94, respectively. Conclusion The proposed Ensemble-GAN framework demonstrated outstanding performance in the semantic segmentation of medical images in comparison with other approaches on popular abdominal imaging benchmarks. The Ensemble-GAN has the potential to segment abdominal images more accurately than human experts.