Lesion Boundary Research Articles

Ultrasound Findings of Suspected Breast Cancer Lesions in Females Accessing Breast Ultrasound Services in Selected Hospitals in Lusaka, Zambia

Background: Breast cancer is the most common malignancy among women in Zambia. Ultrasound can be a useful tool for early diagnosis of breast cancer based on the breast lesions’ features. However, an overlap in the ultrasound features of malignant and benign breast tumours has been noticed together with a paucity of information on the ultrasound features that are more suggestive of breast cancer. Objective: The purpose of this study was to identify ultrasound features of suspected breast cancer lesions which may assist in ascertaining a high breast cancer or benign breast disease suspicion index, that may be used to prioritise patients requiring further and urgent management for breast cancer. Methods: Systematic random sampling was used for the selection of participants in this study. The selected patients first underwent an ultrasound scan in the Ultrasound department to obtain the breast lesion ultrasound features. The histology results for the same patients were then obtained from the University Teaching Hospital (UTH) histology laboratory or the histology report filed in the patient’s medical file. Results: Irregular shape of breast lesions, vertical orientation of breast lesions, hypoechogenicity of breast lesions, complex echo pattern of breast lesions, irregular or spiculated margin contours of breast lesions, compression of breast lesion surrounding tissue, absence of hyperechogenic spots in breast lesions and a breast lesion boundary with neither a hyperechoic halo nor thin capsule (None) were found to be associated with breast cancer. No breast lesion showed less than 3 ultrasound features suspicious for breast cancer. Hyperechoic spots in breast lesions were found to be associated with non-cancerous breast lesions. Conclusion: Low-resource countries, such as Zambia, where access to imaging diagnostic equipment such as Mammography and Magnetic Resonance Imaging machines (MRI), and histology examinations are scarce, the preceding ultrasound findings .......

Stroke-GFCN: ischemic stroke lesion prediction with a fully convolutional graph network.

The interpretation of image data plays a critical role during acute brain stroke diagnosis, and promptly defining the requirement of a surgical intervention will drastically impact the patient's outcome. However, determining stroke lesions purely from images can be a daunting task. Many studies proposed automatic segmentation methods for brain stroke lesions from medical images in different modalities, though heretofore results do not satisfy the requirements to be clinically reliable. We investigate the segmentation of brain stroke lesions using a geometric deep learning model that takes advantage of the intrinsic interconnected diffusion features in a set of multi-modal inputs consisting of computer tomography (CT) perfusion parameters. We propose a geometric deep learning model for the segmentation of ischemic stroke brain lesions that employs spline convolutions and unpooling/pooling operators on graphs to excerpt graph-structured features in a fully convolutional network architecture. In addition, we seek to understand the underlying principles governing the different components of our model. Accordingly, we structure the experiments in two parts: an evaluation of different architecture hyperparameters and a comparison with state-of-the-art methods. The ablation study shows that deeper layers obtain a higher Dice coefficient score (DCS) of up to 0.3654. Comparing different pooling and unpooling methods shows that the best performing unpooling method is the proportional approach, yet it often smooths the segmentation border. Unpooling achieves segmentation results more adapted to the lesion boundary corroborated with systematic lower values of Hausdorff distance. The model performs at the level of state-of-the-art models without optimized training methods, such as augmentation or patches, with a DCS of . We proposed and evaluated an end-to-end trainable fully convolutional graph network architecture using spline convolutional layers for the ischemic stroke lesion prediction. We propose a model that employs graph-based operations to predict acute stroke brain lesions from CT perfusion parameters. Our results prove the feasibility of using geometric deep learning to solve segmentation problems, and our model shows a better performance than other models evaluated. The proposed model achieves improved metric values for the DCS metric, ranging from 8.61% to 69.05%, compared with other models trained under the same conditions. Next, we compare different pooling and unpooling operations in relation to their segmentation results, and we show that the model can produce segmentation outputs that adapt to irregular segmentation boundaries when using simple heuristic unpooling operations.

A collaborative multi-task learning method for BI-RADS category 4 breast lesion segmentation and classification of MRI images

Background and objective: The diagnosis of BI-RADS category 4 breast lesion is difficult because its probability of malignancy ranges from 2% to 95%. For BI-RADS category 4 breast lesions, MRI is one of the prominent noninvasive imaging techniques. In this paper, we research computer algorithms to segment lesions and classify the benign or malignant lesions in MRI images. However, this task is challenging because the BI-RADS category 4 lesions are characterized by irregular shape, imbalanced class, and low contrast.Methods: We fully utilize the intrinsic correlation between segmentation and classification tasks, where accurate segmentation will yield accurate classification results, and classification results will promote better segmentation. Therefore, we propose a collaborative multi-task algorithm (CMTL-SC). Specifically, a preliminary segmentation subnet is designed to identify the boundaries, locations and segmentation masks of lesions; a classification subnet, which combines the information provided by the preliminary segmentation, is designed to achieve benign or malignant classification; a repartition segmentation subnet which aggregates the benign or malignant results, is designed to refine the lesion segment. The three subnets work cooperatively so that the CMTL-SC can identify the lesions better which solves the three challenges.Results and conclusion: We collect MRI data from 248 patients in the Second Hospital of Dalian Medical University. The results show that the lesion boundaries delineated by the CMTL-SC are close to the boundaries delineated by the physicians. Moreover, the CMTL-SC yields better results than the single-task and multi-task state-of-the-art algorithms. Therefore, CMTL-SC can help doctors make precise diagnoses and refine treatments for patients.

Improving adversarial robustness of medical imaging systems via adding global attention noise

Recent studies have found that medical images are vulnerable to adversarial attacks. However, it is difficult to protect medical imaging systems from adversarial examples in that the lesion features of medical images are more complex with high resolution. Therefore, a simple and effective method is needed to address these issues to improve medical imaging systems’ robustness. We find that the attackers generate adversarial perturbations corresponding to the lesion characteristics of different medical image datasets, which can shift the model’s attention to other places. In this paper, we propose global attention noise (GATN) injection, including global noise in the example layer and attention noise in the feature layers. Global noise enhances the lesion features of the medical images, thus keeping the examples away from the sharp areas where the model is vulnerable. The attention noise further locally smooths the model from small perturbations. According to the characteristic of medical image datasets, we introduce Global attention lesion-unrelated noise (GATN-UR) for datasets with unclear lesion boundaries and Global attention lesion-related noise (GATN-R) for datasets with clear lesion boundaries. Extensive experiments on ChestX-ray, Dermatology, and Fundoscopy datasets show that GATN improves the robustness of medical diagnosis models against a variety of powerful attacks and significantly outperforms the existing adversarial defense methods. To be specific, the robust accuracy is 86.66% on ChestX-ray, 72.49% on Dermatology, and 90.17% on Fundoscopy under PGD attack. Under the AA attack, it achieves robust accuracy of 87.70% on ChestX-ray, 66.85% on Dermatology, and 87.83% on Fundoscopy.

On Ultrasonographic Features of Mucinous Carcinoma with Micropapillary Pattern.

To describe the sonographic features of pure mucinous carcinoma with micropapillary pattern (MUMPC) and compare with different pathological type of mucinous breast carcinoma. Subjects were retrospectively reviewed at Suzhou Municipal Hospital from January 2015 to June 2019. Sonographic features of 49 cases (9 MUMPC, 19cPMBC, and 21 MMBC) pathologically confirmed MBC were recorded according to the Breast Imaging Reporting and Data System (BI-RADS) lexicon. The differences in sonographic features among different type of mucinous breast carcinoma were discussed, including clinical features and sonographic features, shape, lesion boundary, peripheral hyperechoic ring, echo pattern, posterior acoustic feature, thickness peripheral hyperechoic area, and blood flow. All MUMPC had no lymph node metastasis (88.9%, 8/9), and most of the MUMPC showed no thickness peripheral hyperechoic area (88.9%, 8/9) and blood flow (55.6%, 5/9) within the tumor. Furthermore, MUMPC had mixed cystic and solid components (33.3%, 3/9) and solid echoic (66.7%, 6/9) structures, with regular shape (66.7%, 6/9) and peripheral hyperechoic ring (66.7%, 6/9). Seven cases of the MUMPC showed circumscribed margin (77.8%, 7/9), and there was significant difference among the three groups (p < 0.05). In addition, there were 7 cases (77.8%, 7/9) of MUMPC tumor ≤2cm, which was significantly different from cPMBC (26.3%, 5/19) and MMBC (28.6%, 6/21) (p < 0.05). There was no significant difference in ultrasonographic features of MBC with different sizes when stratified by tumor size (p > 0.05). Most of the MUMPC showed a circumscribed margin, peripheral hyperechoic ring, and without lymph node metastasis and thickness peripheral hyperechoic area. However, it is challenging to distinguish MUMPC from PMBC and MMBC on ultrasound. Future research should focus on developing novel analysis methods for ultrasound imaging, conducting studies with larger sample sizes and diverse population groups.

The Use of Focused Ultrasound Beams with Shocks to Suppress Diffusion Effects in Volumetric Thermal Ablation of Biological Tissue

The article presents the results of numerical simulation of an experiment on irradiatingex vivobovine liver sample by the therapeutic array of the MR-HIFU clinical system (Sonalleve V1 3.0T, Profound Medical Corp., Canada). Continuous quasi-linear and pulsed shock-wave exposures with the same time-averaged power are compared. Volumetric thermal lesions were generated by moving the focus of the array in its focal plane along discrete trajectories consisting of two or four concentric circles with a maximum radius of 4 mm. The effect of using the criteria for controlling the thermal dose during treatment and ending the sonication on the shape, volume, and exposure time of generating thermal lesion were analyzed. The acoustic field in tissue was calculated using the Westervelt equation; the temperature field was simulated with the inhomogeneous heat conduction equation; and the lesion boundary was determined according to the thermal dose threshold. In the quasi-linear mode corresponding to the clinical one, thermal diffusion leads to elongation of the lesion by a factor of 2–3 along the beam axis compared to the transverse dimension of the trajectory. The use of pulsed shock-wave exposures with switching off the inner circles of the trajectory as they reach the threshold value of the thermal dose makes it possible to significantly suppress the thermal diffusion effects in the axial direction of the beam and obtain localized thermal lesion of a given shape with a thermal ablation rate comparable to the clinical case.

Modeling for Quantitative Analysis of Nakagami Imaging in Accurate Detection and Monitoring of Therapeutic Lesions by High-Intensity Focused Ultrasound.

Nakagami imaging is an appealing monitoring and evaluation technique for high-intensity focused ultrasound treatment when bubbles are present in ultrasound images. This study aimed to investigate the accuracy of thermal lesion detection using Nakagami imaging. Simulations were conducted to explore and quantify the influence of the bubbles and the subresolvable effect at the boundary of the thermal lesion on thermal lesion detection. The thermal ablation experiments were conducted in phantom and porcine liver ex vivo. In the simulation, the estimated lateral and axial size of the thermal lesion in the Nakagami image was 4.91 and 4.79 mm, close to the actual size (5×5 mm). The simulation results indicated that the subresolvable region in high-intensity focused ultrasound treatment thermal ablation mainly happened at the boundary between bubbles and the untreated region and does not affect the accuracy of thermal lesion detection. The accurate detection of the thermal lesion using Nakagami imaging mainly depends on bubbles and thermal lesion characterization. Our thermal ablation experiments confirmed that Nakagami imaging has the ability to accurately identify thermal lesions from bubbles. The subresolvable effect is helpful for thermal lesion identification, and precision is related to the Nakagami values chosen for boundary division in Nakagami imaging. Therefore, Nakagami imaging is a promising method for accurately evaluating thermal lesions. Further studies in vivo and in clinical settings will be needed to explore its potential applications.

Effective hybrid attention network based on pseudo-color enhancement in ultrasound image segmentation

Ultrasound image segmentation plays a vital role in the early diagnosis of human diseases. It helps diagnose many diseases, such as breast cancer, hemangioma, and other gynecological disorders. However, the intrinsic imaging characteristics of ultrasound images result in substantially lower resolution and clarity than CT, MRI, and other imaging modalities, and they are sensitive to interference from external influences. With its inherent artifacts, blurred lesion boundaries, and uneven intensity distribution, ultrasound images present a challenging task when it comes to segmenting lesion areas accurately. In recent years, convolutional neural networks (CNNs) have achieved remarkable results in medical image segmentation tasks. However, CNNs are limited in capturing the remote dependencies of the input image, leading to degraded accuracy in segmenting ultrasound lesions. In this paper, we developed a deep convolutional neural network that incorporates the pseudo-color enhancement algorithm and hybrid attention modules that enhance the network’s ability to extract fine features and remote modeling capabilities. We propose a novel multi-scale channel attention-based decoder that efficiently uses the feature maps from the encoder as a complement and fuses them with the upsampled feature maps. The hybrid attention combination captures cross-channel interactions efficiently and enhances the context modeling capability, further improving the extraction of coarse and delicate features, and resulting in significant performance improvements. We found that the dice performance improved by 2.54%, 2.47%, 1.39%, 0.99%, and 1.23% on the BUL, BUSI, Hemangioma, BP, and VUI. Results from four public datasets and one self-collected dataset indicate that the proposed method outperforms other medical image segmentation methods for ultrasound image lesion segmentation.

Medical image mis-segmentation region refinement framework based on dynamic graph convolution

In medical image segmentation tasks, it is hard for traditional Convolutional Neural Network (CNN) to capture essential information such as spatial structure and global contextual semantic features since it suffers from a limited receptive field. The deficiency weakens the CNN segmentation performance in the lesion boundary regions. To handle the aforementioned problems, a medical image mis-segmentation region refinement framework based on dynamic graph convolution is proposed to refine the boundary and under-segmentation regions. The proposed framework first employs a lightweight dual-path network to detect the boundaries and nearby regions, which can further obtain potentially misclassified pixels from the coarse segmentation results of the CNN. Then, we construct the pixels into the appropriate graphs by CNN-extracted features. Finally, we design a dynamic residual graph convolutional network to reclassify the graph nodes and generate the final refinement results. We chose UNet and its eight representative improved networks as the basic networks and tested them on the COVID, DSB, and BUSI datasets. Experiments demonstrated that the average Dice of our framework is improved by 1.79%, 2.29%, and 2.24%, the average IoU is improved by 2.30%, 3.53%, and 2.39%, and the Se is improved by 5.08%, 4.78%, and 5.31% respectively. The experimental results prove that the proposed framework has the refinement capability to remarkably strengthen the segmentation result of the basic network. Furthermore, the framework has the advantage of high portability and usability, which can be inserted into the end of mainstream medical image segmentation networks as a plug-and-play enhancement block.

The Value of Contrast-Enhanced Ultrasound Features in Predicting HER2 Positive and Ki67 High Expression in Invasive Breast Cancer of No Special Type

Objective: To investigate the value of contrast-enhanced ultrasound features in predicting HER2 positive and Ki67 high expression in invasive breast cancer of no special type. Methods: Retrospective analysis was conducted on the correlation between contrast-enhanced ultrasound images of 92 patients with invasive breast cancer of no special type confirmed by pathology in our hospital and the expression of HER2 and Ki67, including 64 cases in the HER2 negative group and 28 cases in the positive group. Ki67 expression was low in 14 cases and high in 78 cases. The contrast-enhanced ultrasound indicators include: BI (Base Intensity), AT (Arrival Time), TTP (Time To Peak), PI (Peak Intensity), AS (Ascending Slope), DT/2 (Descending Time/2), DS (Descending Slope), AUC (Area Under Curve), maximum focal area ratio of before and after contrast-enhanced ultrasound (AR Area Ratio), Distance of anterior boundary of lesion to skin after contrast-enhanced ultrasound (DTS Distance to skin) and maximum diameter after contrast-enhanced ultrasound (D Diameter), and the above indexes were compared and analyzed in the negative and positive groups of HER2, and the low and high expression groups of Ki67. Results: Logistics regression analysis showed that DTS is an independent risk factor for HER2 positive (P＜0. 05). AR was an independent risk factor for Ki67 high expression (P＜0. 05). Conclusion: The combination of contra-enhanced ultrasound features has high diagnostic efficacy for HER2 positive and Ki67 high expression, which can provide an important reference.

Cross-image Dependency Modelling for Breast Ultrasound Segmentation.

We present a novel deep network (namely BUSSeg) equipped with both within- and cross-image long-range dependency modeling for automated lesions segmentation from breast ultrasound images, which is a quite daunting task due to (1) the large variation of breast lesions, (2) the ambiguous lesion boundaries, and (3) the existence of speckle noise and artifacts in ultrasound images. Our work is motivated by the fact that most existing methods only focus on modeling the within-image dependencies while neglecting the cross-image dependencies, which are essential for this task under limited training data and noise. We first propose a novel cross-image dependency module (CDM) with a cross-image contextual modeling scheme and a cross-image dependency loss (CDL) to capture more consistent feature expression and alleviate noise interference. Compared with existing cross-image methods, the proposed CDM has two merits. First, we utilize more complete spatial features instead of commonly used discrete pixel vectors to capture the semantic dependencies between images, mitigating the negative effects of speckle noise and making the acquired features more representative. Second, the proposed CDM includes both intra- and inter-class contextual modeling rather than just extracting homogeneous contextual dependencies. Furthermore, we develop a parallel bi-encoder architecture (PBA) to tame a Transformer and a convolutional neural network to enhance BUSSeg's capability in capturing within-image long-range dependencies and hence offer richer features for CDM. We conducted extensive experiments on two representative public breast ultrasound datasets, and the results demonstrate that the proposed BUSSeg consistently outperforms state-of-the-art approaches in most metrics.

Dynamically aggregating MLPs and CNNs for skin lesion segmentation with geometry regularization

Background and objectiveMelanoma is a highly malignant skin tumor. Accurate segmentation of skin lesions from dermoscopy images is pivotal for computer-aided diagnosis of melanoma. However, blurred lesion boundaries, variable lesion shapes, and other interference factors pose a challenge in this regard. MethodsThis work proposes a novel framework called CFF-Net (Cross Feature Fusion Network) for supervised skin lesion segmentation. The encoder of the network includes dual branches, where the CNNs branch aims to extract rich local features while MLPs branch is used to establish both the global-spatial-dependencies and global-channel-dependencies for precise delineation of skin lesions. Besides, a feature-interaction module between two branches is designed for strengthening the feature representation by allowing dynamic exchange of spatial and channel information, so as to retain more spatial details and inhibit irrelevant noise. Moreover, an auxiliary prediction task is introduced to learn the global geometric information, highlighting the boundary of the skin lesion. ResultsComprehensive experiments using four publicly available skin lesion datasets (i.e., ISIC 2018, ISIC 2017, ISIC 2016, and PH2) indicated that CFF-Net outperformed the state-of-the-art models. In particular, CFF-Net greatly increased the average Jaccard Index score from 79.71% to 81.86% in ISIC 2018, from 78.03% to 80.21% in ISIC 2017, from 82.58% to 85.38% in ISIC 2016, and from 84.18% to 89.71% in PH2 compared with U-Net. Ablation studies demonstrated the effectiveness of each proposed component. Cross-validation experiments in ISIC 2018 and PH2 datasets verified the generalizability of CFF-Net under different skin lesion data distributions. Finally, comparison experiments using three public datasets demonstrated the superior performance of our model. ConclusionThe proposed CFF-Net performed well in four public skin lesion datasets, especially for challenging cases with blurred edges of skin lesions and low contrast between skin lesions and background. CFF-Net can be employed for other segmentation tasks with better prediction and more accurate delineation of boundaries.

Dual-feature Fusion Attention Network for Small Object Segmentation

Accurate segmentation of medical images is an important step during radiotherapy planning and clinical diagnosis. However, manually marking organ or lesion boundaries is tedious, time-consuming, and prone to error due to subjective variability of radiologist. Automatic segmentation remains a challenging task owing to the variation (in shape and size) across subjects. Moreover, existing convolutional neural networks based methods perform poorly in small medical objects segmentation due to class imbalance and boundary ambiguity. In this paper, we propose a dual feature fusion attention network (DFF-Net) to improve the segmentation accuracy of small objects. It mainly includes two core modules: the dual-branch feature fusion module (DFFM) and the reverse attention context module (RACM). We first extract multi-resolution features by multi-scale feature extractor, then construct DFFM to aggregate the global and local contextual information to achieve information complementarity among features, which provides sufficient guidance for accurate small objects segmentation. Moreover, to alleviate the degradation of segmentation accuracy caused by blurred medical image boundaries, we propose RACM to enhance the edge texture of features. Experimental results on datasets NPC, ACDC, and Polyp demonstrate that our proposed method has fewer parameters, faster inference, and lower model complexity, and achieves better accuracy than more state-of-the-art methods.

Geographic atrophy: Mechanism of disease, pathophysiology, and role of the complement system.

Geographic atrophy (GA) is an advanced form of age-related macular degeneration (AMD), characterized by atrophic lesions that first start in the outer retina and progressively expand to cover the macula and the fovea, the center of the macula, leading to irreversible loss of vision over time. GA is distinct from wet or neovascular AMD (nAMD), the other form of advanced AMD. Neovascular AMD is characterized by new invading leaky blood vessels in the macula that can lead to acute vision loss. GA and nAMD may coexist in the same eye. The underlying pathophysiology of GA is complex and thought to involve chronic inflammation due to overactivation of the complement system that leads to the loss of photoreceptors, retinal pigment epithelium (RPE), and the underlying choriocapillaris. The disappearance of these structures appears as sharply demarcated atrophic lesions that are typical of GA. Researchers have reported about 1 million reported cases of GA in the United States, and about 160,000 cases occur per year. The most important risk factors for GA are increasing age and family history. Diagnosis of GA is usually made by using multimodal imaging techniques. Lesions associated with GA are highly heterogeneous, and the growth rate may differ from patient to patient. Despite the progressive nature of GA, the fovea may be spared until much later in the disease, thereby retaining central vision in patients. With time, atrophic lesions may progressively grow to involve the fovea, thereby severely impairing central vision. Vision loss can happen rapidly once the lesions reach the fovea. However, even without the involvement of the fovea, ongoing vision impairment impacting daily life may be present. Median time from GA not involving the center of the fovea (without subfoveal involvement) to GA with lesion boundary affecting the foveal center (subfoveal involvement) ranges from 1.4 to 2.5 years. GA can greatly impact patients' functioning and quality of life and limit their independence by interfering with activities of daily living, including difficulties with reading, driving, watching television, recognizing faces, and being unable to do household chores. No treatments have been available until intravitreal pegcetacoplan was recently approved by the US Food and Drug Administration for GA secondary to AMD. DISCLOSURES: Dr Bakri serves as a consultant to Apellis Pharmaceuticals, as well as AbbVie, Adverum, Eyepoint, iLumen, Iveric Bio, Genentech, Novartis, Outlook Therapeutics, Pixium, Regeneron, Roche, and Regenxbio. Drs Sharp, Luo, and Sarda are employees of Apellis Pharmaceuticals. Dr Bektas and Ms Khan are employees of RTI Health Solutions. Apellis developed and led the concept design of this publication, review and interpretation, approval, and decision to publish. This research was developed under a research contract between RTI Health Solutions and Apellis Pharmaceuticals and was funded by Apellis Pharmaceuticals. This supplement is to describe the disease of geographic atrophy and was funded by Apellis. Apellis Pharmaceuticals has developed Syfovre (pegcetacoplan), the first and only treatment for geographic atrophy.

DIAGNOSTIC ACCURACY OF CONTRAST-ENHANCED ULTRASOUND TO DIFFERENTIATE BENIGN AND MALIGNANT BREAST LESIONS

Cancer of the breast is caused by a cluster of abnormal breast cells that continue to divide and spread throughout the breast. These cells eventually aggregate as a lump in the affected breast. If the cancerous tumor is not treated, the cancer cells may spread to other areas of the body through a process known as metastasis. A metastasis may develop in the lymph nodes located in the axilla or those located above the scapula. In addition, cancer cells have the ability to colonize the bones, lungs, liver, skin, and even the space directly under the skin. Cancer of the breast tissue is a malignancy that can start from either the ductal or lobule epithelium of the breast. Breast cancer is the most common kind of cancer in women. The ability of ultrasonography to identify cystic masses is one of the technology's many strengths. According to BIRADS, the American College of Radiology has also developed a standardized language for the reading and reporting of ultrasounds, which is quite similar to the language used for mammography. Shape of the mass, borders, orientation, kind of posterior acoustic, lesion boundaries, and echo pattern were the characteristics that were described. CEUS is an imaging technology that evaluates breast lesions by analyzing their qualitative qualities as well as their quantitative aspects. The detection of lymphatic channels and sentinel lymph nodes can be done using this method, and it is practical and effective.

Progressive growing of Generative Adversarial Networks for improving data augmentation and skin cancer diagnosis

Early melanoma diagnosis is the most important factor in the treatment of skin cancer and can effectively reduce mortality rates. Recently, Generative Adversarial Networks have been used to augment data, prevent overfitting and improve the diagnostic capacity of models. However, its application remains a challenging task due to the high levels of inter and intra-class variance seen in skin images, limited amounts of data, and model instability. We present a more robust Progressive Growing of Adversarial Networks based on residual learning, which is highly recommended to ease the training of deep networks. The stability of the training process was increased by receiving additional inputs from preceding blocks. The architecture is able to produce plausible photorealistic synthetic 512 × 512 skin images, even with small dermoscopic and non-dermoscopic skin image datasets as problem domains. In this manner, we tackle the lack of data and the imbalance problems. Additionally, the proposed approach leverages a skin lesion boundary segmentation algorithm and transfer learning to enhance the diagnosis of melanoma. Inception score and Matthews Correlation Coefficient were used to measure the performance of the models. The architecture was evaluated qualitatively and quantitatively through the use of an extensive experimental study on sixteen datasets, illustrating its effectiveness in the diagnosis of melanoma. Finally, four state-of-the-art data augmentation techniques applied in five convolutional neural network models were significantly outperformed. The results indicated that a bigger number of trainable parameters will not necessarily obtain a better performance in melanoma diagnosis.

Discrimination tasks in simulated low-dose CT noise.

This study reports the results of a set of discrimination experiments using simulated images that represent the appearance of subtle lesions in low-dose computed tomography (CT) of the lungs. Noise in these images has a characteristic ramp-spectrum before apodization by noise control filters. We consider three specific diagnostic features that determine whether a lesion is considered malignant or benign, two system-resolution levels, and four apodization levels for a total of 24 experimental conditions. The goal of the investigation is to better understand how well human observers perform subtle discrimination tasks like these, and the mechanisms of that performance. We use a forced-choice psychophysical paradigm to estimate observer efficiency and classification images. These measures quantify how effectively subjects can read the images, and how they use images to perform discrimination tasks across the different imaging conditions. The simulated CT images used as stimuli in the psychophysical experiments are generated from high-resolution objects passed through a modulation transfer function (MTF) before down-sampling to the image-pixel grid. Acquisition noise is then added with a ramp noise-power spectrum (NPS), with subsequent smoothing through apodization filters. The features considered are lesion size, indistinct lesion boundary, and a nonuniform lesion interior. System resolution is implemented by an MTF with resolution (10% max.) of 0.47 or 0.58 cyc/mm. Apodization is implemented by a Shepp-Logan filter (Sinc profile) with various cutoffs. Six medically naïve subjects participated in the psychophysical studies, entailing training and testing components for each condition. Training consisted of staircase procedures to find the 80% correct threshold for each subject, and testing involved 2000 psychophysical trials at the threshold value for each subject. Human-observer performance is compared to the Ideal Observer to generate estimates of task efficiency. The significance of imaging factors is assessed using ANOVA. Classification images are used to estimate the linear template weights used by subjects to perform these tasks. Classification-image spectra are used to analyze subject weights in the spatial-frequency domain. Overall, average observer efficiency is relatively low in these experiments (10%-40%) relative to detection and localization studies reported previously. We find significant effects for feature type and apodization level on observer efficiency. Somewhat surprisingly, system resolution is not a significant factor. Efficiency effects of the different features appear to be well explained by the profile of the linear templates in the classification images. Increasingly strong apodization is found to both increase the classification-image weights and to increase the mean-frequency of the classification-image spectra. A secondary analysis of "Unapodized" classification images shows that this is largely due to observers undoing (inverting) the effects of apodization filters. These studies demonstrate that human observers can be relatively inefficient at feature-discrimination tasks in ramp-spectrum noise. Observers appear to be adapting to frequency suppression implemented in apodization filters, but there are residual effects that are not explained by spatial weighting patterns. The studies also suggest that the mechanisms for improving performance through the application of noise-control filters may require further investigation.

JAAL-Net: a joint attention and adversarial learning network for skin lesion segmentation

Objective. Skin lesion segmentation plays an important role in the diagnosis and treatment of melanoma. Existing skin lesion segmentation methods have trouble distinguishing hairs, air bubbles, and blood vessels around lesions, which affects the segmentation performance. Approach. To clarify the lesion boundary and raise the accuracy of skin lesion segmentation, a joint attention and adversarial learning network (JAAL-Net) is proposed that consists of a generator and a discriminator. In the JAAL-Net, the generator is a local fusion network (LF-Net) utilizing the encoder-decoder structure. The encoder contains a convolutional block attention module to increase the weight of lesion information. The decoder involves a contour attention to obtain edge information and locate the lesion. To aid the LF-Net generate higher confidence predictions, a discriminant dual attention network is constructed with channel attention and position attention. Main results. The JAAL-Net is evaluated on three datasets ISBI2016, ISBI2017 and ISIC2018. The intersection over union of the JAAL-Net on the three datasets are 90.27%, 89.56% and 80.76%, respectively. Experimental results show that the JAAL-Net obtains rich lesion and boundary information, enhances the confidence of the predictions, and improves the accuracy of skin lesion segmentation. Significance. The proposed approach effectively improves the performance of the model for skin lesion segmentation, which can assist physicians in accurate diagnosis well.

RMAU-Net: Residual Multi-Scale Attention U-Net For liver and tumor segmentation in CT images

Liver cancer is one of the leading causes of cancer-related deaths worldwide. Automatic liver and tumor segmentation are of great value in clinical practice as they can reduce surgeons’ workload and increase the probability of success in surgery. Liver and tumor segmentation is a challenging task because of the different sizes, shapes, blurred boundaries of livers and lesions, and low-intensity contrast between organs within patients. To address the problem of fuzzy livers and small tumors, we propose a novel Residual Multi-scale Attention U-Net (RMAU-Net) for liver and tumor segmentation by introducing two modules, i.e., Res-SE-Block and MAB. The Res-SE-Block can mitigate the problem of gradient disappearance by residual connection and enhance the quality of representations by explicitly modeling the interdependencies and feature recalibration between the channels of features. The MAB can exploit rich multi-scale feature information and capture inter-channel and inter-spatial relationships of features simultaneously. In addition, a hybrid loss function, that combines focal loss and dice loss, is designed to improve segmentation accuracy and speed up convergence. We evaluated the proposed method on two publicly available datasets, i.e., LiTS and 3D-IRCADb. Our proposed method achieved better performance than the other state-of-the-art methods, with dice scores of 0.9552 and 0.9697 for LiTS and 3D-IRCABb liver segmentation, and dice scores of 0.7616 and 0.8307 for LiTS and 3D-IRCABb liver tumor segmentation.

Evaluation of the predictive values of collapse and necrotic lesion boundary for osteonecrosis of the femoral head prognosis.

Osteonecrosis of the femoral head (ONFH) is a disabling and intractable orthopedic disease largely affecting young and middle-aged groups. Current standard of treatment relies on the collapse of femoral head as a predictor for prognosis. However, a wide range of variability in repair potentials is observed in patients with femoral head collapse. Therefore, the present study aimed to evaluate the accuracy of femoral head collapse as a predictor and to propose the necrotic lesion boundary as a novel yet reliable measure for ONFH prognosis. A retrospective cross-sectional study was conducted at the First Affiliated Hospital of Guangzhou University of Chinese Medicine, 203 hips with ONFH from 134 patients were included. The occurrences and progression of femoral head collapse were recorded. Necrosis lesion boundary was quantified and classified for each case based on anteroposterior view intact ratio (APIR) and the frog-leg view intact ratio (FLIR) as independent variables. Dependent variables were defined as progressive collapse or terminal collapse for Association Research Circulation Osseous (ARCO) stage II and III respectively. Logistic regression analysis, Receiver Operating Characteristic (ROC) curve and Kaplan-Meier (K-M) survival analysis was performed and results were interpreted. Out of the 106 hips in ARCO stage II, 31 hips collapsed with further progression, while 75 hips had no collapse or collapse with repair of the necrotic areas. Out of the 97 hips in ARCO stage IIIA, the collapse continued to progress in 58 hips while the necrotic areas were repaired in 39 hips. Logistic regression analysis demonstrated that both APIR and FLIR, were independent risk factors. Further ROC curve analysis indicated that the cutoff values of APIR and FLIR could be considered as indications for evaluating the prognosis of ONFH. Contrary to the traditional view of poor prognosis after femoral head collapse, K-M survival analysis demonstrated a high value of APIR and FLIR for ONFH prognosis. The present study found that the occurrence of collapse is an oversimplified predictor for ONFH prognosis. The collapse of the femoral head in ONFH does not predict a poor prognosis. The necrosis lesion boundary has a high value in predicting ONFH prognosis and informing clinical treatment strategies.