Lesion Boundary Research Articles

Speckle Reduction via Deep Content-Aware Image Prior for Precise Breast Tumor Segmentation in an Ultrasound Image.

The performance of computer-aided diagnosis (CAD) systems that are based on ultrasound imaging has been enhanced owing to the advancement in deep learning. However, because of the inherent speckle noise in ultrasound images, the ambiguous boundaries of lesions deteriorate and are difficult to distinguish, resulting in the performance degradation of CAD. Although several methods have been proposed to reduce speckle noise over decades, this task remains a challenge that must be improved to enhance the performance of CAD. In this article, we propose a deep content-aware image prior (DCAIP) with a content-aware attention module (CAAM) for superior despeckling of ultrasound images without clean images. For the image prior, we developed a CAAM to deal with the content information in an input image. In this module, super-pixel pooling (SPP) is used to give attention to salient regions in an ultrasound image. Therefore, it can provide more content information regarding the input image when compared to other attention modules. The DCAIP consists of deep learning networks based on this attention module. The DCAIP is validated by applying it as a preprocessing step for breast tumor segmentation in ultrasound images, which is one of the tasks in CAD. Our method improved the segmentation performance by 15.89% in terms of the area under the precision-recall (PR) curve (AUPRC). The results demonstrate that our method enhances the quality of ultrasound images by effectively reducing speckle noise while preserving important information in the image, promising for the design of superior CAD systems.

Mutual gain adaptive network for segmenting brain stroke lesions

Brain stroke, an acute vascular disease, can lead to brain damage. As the incidence rate continues to increase, it becomes urgent and yet significantly to develop automatic tools for segmenting brain stroke lesions, for helping radiologists to detect brain stroke both effectively and accurately. Deep learning models, though have advanced this task, they are still suffering from the huge size variation and ambiguous boundaries of stroke lesions, both substantially degrading the segmentation performance. In this paper, we proposed an effective and generic learning network, called mutual gain adaptive network, to enhance the network’s ability on dealing with size variation and distinguishing ambiguous boundaries, for advancing deep neural networks in brain stroke lesion segmentation. The main contribution is that 1) a mutual gain adaptive similarity (MGAS) block and 2) a global context-awareness (GCA) block. MGAS block aims at capturing both short- and long-range spatial dependencies, for better encoding global contexts by leveraging feature similarity, to enhance feature activation for segmenting stroke lesions with different sizes. GCA block is designed to supply global context for low-semantic features, for facilitating the restoration of pixel localization to distinguish ambiguous boundaries. Our network is assessed in a publicly available dataset, e.g. Anatomical Tracings of Lesions After Stroke (ATLAS). Extensive experimental results show that it is better than state-of-the-art methods.

Clinicopathological features of early gastric cancer after Helicobacter pylori eradication

Objective: To investigate the clinicopathological features of early gastric cancers after Helicobacter pylori (H. pylori) eradication. Methods: The clinical data of 26 cases of gastric cancer that were diagnosed after H. pylori eradication and 45 cases without H. pylori eradication in the 989 Hospital of the Joint Logistics Support Force of the People's Liberation Army (the former 152 Hospital), Pingdingshan, China from 2013 to 2021 were collected. The histological, immunophenotypic and clinical characteristics of the two groups were compared, and discussed with review of the related literature. Results: Among the gastric cancer patients with H. pylori eradication, there were 20 males and 6 females with a median age of 65 years (range 53 to 77 years). The cancer involved the upper part of the stomach in 12 cases, the middle part of the stomach in 4 cases, and the lower part of the stomach in 10 cases. The median diameter of the tumors was 12 mm (range 4-29 mm). According to the Paris Classification, 4 cases were 0-Ⅱa, 4 cases were 0-Ⅱb, 18 cases were 0-Ⅱc. White light endoscopy showed that the lesions were reddish to yellowish. The lesion boundary was clear in 12 cases and was unclear or gastritis-like changes in 14 cases, while the irregular microvascular structure and microsurface structure, as well as the relatively visible spinous boundary, were visible under narrow-band imaging. There were 20 cases of well-differentiated tubular adenocarcinoma, 4 cases of highly to moderately differentiated tubular adenocarcinoma, and 2 cases of well-differentiated tubular adenocarcinoma with papillary adenocarcinoma. Compared with gastric cancers without H. pylori eradication, gastric cancers diagnosed after H. pylori eradication was associated with lower nucleus-cytoplasm ratio (<50%), normal epithelial coverage on the cancer surface, mild atypical epithelial coverage on the cancer surface, elongation of non-cancerous glands in the cancer tissue and subepithelial progression of cancerous glands were higher (P<0.05). The cellular immunophenotypes were gastric type in 6 cases, intestinal type in 4 cases and gastrointestinal mixed type in 16 cases. Conclusions: The early gastric cancers diagnosed after H. pylori eradication are more subtle clinically and mostly well-differentiated tubular adenocarcinoma. The important morphological features of gastric cancer diagnosed after H. pylori eradication are decreased cytological atypia and overlying normal epithelium or mildly atypical epithelium of the cancer. Understanding and recognizing these morphological features are helpful to make correct endoscopic and pathological diagnoses.

Comparison of High-Frequency Contrast-Enhanced Ultrasound With Conventional High-Frequency Ultrasound in Guiding Pleural Lesion Biopsy.

The goal of the work described here was to compare high-frequency contrast-enhanced ultrasound (HF-CEUS) and conventional high-frequency ultrasound (HFU) with respect to performance and safety during ultrasound-guided biopsy of pleural lesions. We performed a retrospective study on patients with pleural lesions who received the puncture biopsy under the guidance of conventional HFU or HF-CEUS between August 2018 and August 2021. These patients received either a conventional HF-U (HF-U group) or HF-CEUS (HF-CEUS group) examination. Clinical characteristics, pathological results, ultrasonic images and complications were compared between these two groups. A total of 144 patients were enrolled, with 70 in the HFU group and 74 in the HF-ECUS group. Except for the time required for ultrasonic localization (p < 0.05), there were no significant differences in clinical characteristics between the two groups. The success rate of biopsy in the HF-CEUS group was higher than that in the HFU group (93.2% vs. 81.4%, p < 0.05). There were significant differences between the two groups in terms of measurements of pleural thickness, necrotic areas, large blood vessels and lesion boundaries (p < 0.05). The complication rates were 2.7% and 12.9% in the HF-CEUS and HFU groups, respectively, with a significant difference (p < 0.05). Compared with conventional HFU, the HF-CEUS-guided pleural biopsy had a better success rate and fewer complications. HF-CEUS could facilitate the biopsy in patients with pleural lesions.

A Novel Approach for the Shape Characterisation of Non-Melanoma Skin Lesions Using Elliptic Fourier Analyses and Clinical Images.

The early detection of Non-Melanoma Skin Cancer (NMSC) is crucial to achieve the best treatment outcomes. Shape is considered one of the main parameters taken for the detection of some types of skin cancer such as melanoma. For NMSC, the importance of shape as a visual detection parameter is not well-studied. A dataset of 993 standard camera images containing different types of NMSC and benign skin lesions was analysed. For each image, the lesion boundaries were extracted. After an alignment and scaling, Elliptic Fourier Analysis (EFA) coefficients were calculated for the boundary of each lesion. The asymmetry of lesions was also calculated. Then, multivariate statistics were employed for dimensionality reduction and finally computational learning classification was employed to evaluate the separability of the classes. The separation between malignant and benign samples was successful in most cases. The best-performing approach was the combination of EFA coefficients and asymmetry. The combination of EFA and asymmetry resulted in a balanced accuracy of 0.786 and an Area Under Curve of 0.735. The combination of EFA and asymmetry for lesion classification resulted in notable success rates when distinguishing between benign and malignant lesions. In light of these results, skin lesions’ shape should be integrated as a fundamental part of future detection techniques in clinical screening.

Dual-modality fibre optic probe for simultaneous ablation and ultrasound imaging

All-optical ultrasound (OpUS) is an emerging high resolution imaging paradigm utilising optical fibres. This allows both therapeutic and imaging modalities to be integrated into devices with dimensions small enough for minimally invasive surgical applications. Here we report a dual-modality fibre optic probe that synchronously performs laser ablation and real-time all-optical ultrasound imaging for ablation monitoring. The device comprises three optical fibres: one each for transmission and reception of ultrasound, and one for the delivery of laser light for ablation. The total device diameter is < 1 mm. Ablation monitoring was carried out on porcine liver and heart tissue ex vivo with ablation depth tracked using all-optical M-mode ultrasound imaging and lesion boundary identification using a segmentation algorithm. Ablation depths up to 2.1 mm were visualised with a good correspondence between the ultrasound depth measurements and visual inspection of the lesions using stereomicroscopy. This work demonstrates the potential for OpUS probes to guide minimally invasive ablation procedures in real time.

304 A residual dense network for skin lesion segmentation in dermoscopy images

Skin lesion segmentation in dermoscopic images is more challenging due to the irregular and blurred skin lesion boundaries, as well as the low visual contrast between the skin lesions and the surrounding normal tissues. This paper proposes a residual dense network for skin lesion segmentation in dermoscopy images. Compared with the existing mainstream image segmentation methods, we propose a novel residual dense module and increase the depth of the convolutional neural network, which not only makes the network easier to converge, but also enables each layer of the network to obtain sufficient information interaction.

LB1005 Skin marking for reflectance confocal microscopy

Reflectance confocal microscopy (RCM) generates serial depth images with either 8mm mosaic or 0.6mm field views with the Vivascope 1500 or 3000, respectively. Using 830nm light produces greyscale cellular level images of features that refract light at that narrow portion of the spectrum. Margin mapping for lentigo maligna and lentigo maligna melanoma with RCM can be challenging, especially without clear lesion boundaries or visible markings within confocal images. In some cases, orientation of anatomical direction is important, while in other cases, marking the boundaries of a particular finding is critical.

Radiological Features of IgG4-Related Disease in the Head, Neck, and Skull Base

Background: IgG4-related disease (IgG4-RD) is increasingly recognized as a multisystemic, chronic inflammatory process characterized by histologic fibrosis with IgG4-positive plasma cell infiltration. Objectives: The purpose of this study was to characterize the imaging features of patients diagnosed with IgG4-RD in the head and neck, especially the skull base. Methods: Our study evaluated CT and MR imaging features of IgG4-RD in the head, neck, and skull base. Images from 15 patients were retrospectively evaluated for the location, signal intensity, morphology, size, boundary, and pre- and post-contrast MRI performances. Results: The lesions presented as irregular shaped, localized masses, distributed in skull base regions; 93.3% of the lesions were isointensity in T1WI (14/15). A total of 80% of the lesions were iso-hypointense in T2WI (12/15); 60% of the lesions got homogeneous enhancement (9/15); and 46.7% of the patients had cranial nerves dysfunction (7/15). The most likely involved cranial nerve was trigeminal nerves (5/15); 60% of the patients had osteolytic bone destruction or sclerosis (9/15). Conclusion: Typical radiological features of IgG4-RD included T1 isointensity and T2 hypointensity, homogeneous and gradual enhancement pattern in MRI, easy cranial nerve invasion, dura involvement but the absence of brain edema, and the presence of bone remodeling without destruction, blurred lesion boundaries.

Deep learning methods for lesion detection on mammography images: a comparative analysis.

Automatic lesion segmentation in mammography images assists in the diagnosis of breast cancer, which is the most common type of cancer especially among women. The robust segmentation of mammography images has been considered a backbreaking task due to: i) the low contrast of the lesion boundaries; ii) the extremely variable lesions' sizes and shapes; and iii) some extremely small lesions on the mammogram image. To overcome these drawbacks, Deep Learning methods have been implemented and have shown impressive results when applied to medical image segmentation. This work presents a benchmark for breast lesion segmentation in mammography images, where six state-of-the-art methods were evaluated on 1692 mammograms from a public dataset (CBIS-DDSM), and compared considering the following six metrics: i) Dice coefficient; ii) Jaccard index; iii) accuracy; iv) recall; v) specificity; and vi) precision. The base U-Net, UNETR, DynUNet, SegResNetVAE, RF-Net, MDA-Net architectures were trained with a combination of the cross-entropy and Dice loss functions. Although the networks presented Dice scores superior to 86%, two of them managed to distinguish themselves. In general, the results demonstrate the efficiency of the MDA-Net and DynUnet with Dice scores of 90.25% and 89.67%, and accuracy of 93.48% and 93.03%, respectively. Clinical Relevance--- The presented comparative study allowed to identify the current performance of deep learning strategies on the segmentation of breast lesions.

Focal U-Net: A Focal Self-attention based U-Net for Breast Lesion Segmentation in Ultrasound Images.

Accurate breast lesion segmentation in ultrasound images helps radiologists to make exact diagnoses and treatments, which is important to increase the survival rate of breast cancer patients. Recently, deep learning-based methods have demonstrated remarkable results in breast lesion segmentation. However, the blurry breast lesion boundaries and noise artifacts in ultrasound images still limit the performance of the deep learning-based methods. In this paper, we propose a novel segmentation network equipped with a focal self-attention block for improving the performance of breast lesion segmentation. The focal self-attention block can incorporate fine-grained local and coarse-grained global information. The fine-grained local information is useful to enhance features of breast lesion boundaries, while the coarse-grained global information effectively reduces noise interference. To verify the performance of our network, we implement breast lesion segmentation on our collected dataset of 9836 ultrasound images. The results demonstrate that the focal self-attention block enhances features of breast lesion boundaries and improves the accuracy of breast lesion segmentation.

A Novel Surgical Method for Congenital Lung Malformations: A Pilot Study

Thoracoscopic surgery for congenital lung malformations (CLMs) has been widely used. However, there are still many shortcomings in the current surgical methods, such as the removal of some normal lung tissue and residual lesions. Thoracoscopic anatomical lesion resection (TALR) may be an effective and safe method to resolve these problems. This retrospective study was conducted to estimate the safety and efficiency of TALR. A retrospective review of clinical data involving patients who underwent TALR in our hospital from October 2019 to January 2021 was performed. Clinical data, including patients' demographic characteristics, manipulative details, and postoperative complications, were extracted and analyzed. A total of 95 cases were included in this study. All cases were operated on under thoracoscope, with no conversion to open surgery. The average operation time was 63.2 ± 15.2 min (range 36-142 min). The average bleeding volume during the operation was 5.8 ± 2.1 mL (range 2-10 mL). The mean diameter of the lesion was 4.4 ± 1.9 cm (range 3-7 cm). All cases had no complications, such as bronchopleurothelial fistula, hemorrhage, atelectasis, or pulmonary infection. Two patients developed pneumothorax 1 month after the operation and were cured by closed thoracic drainage. One month after the operation, chest CT showed regular cysts in 2 cases. The other patients were followed up for over 1 year by chest CT after the operation, and all recovered well without residual lesions or recurrence. Combined with the preoperative thin slice CT and intraoperative lesion boundary, anatomical intrapulmonary dissociation based on the pulmonary vein system, TALR, as a lung-sparing surgery of completely removing the lesion and retaining all normal lung tissue, has high safety and effectiveness in the treatment of CLMs.

Precision Navigation of Hepatic Ischemia-Reperfusion Injury Guided by Lysosomal Viscosity-Activatable NIR-II Fluorescence.

Hepatic ischemia–reperfusion injury (HIRI) is responsible for postoperative liver dysfunction and liver failure. Precise and rapid navigation of HIRI lesions is critical for early warning and timely development of pretreatment plans. Available methods for assaying liver injury fail to provide the exact location of lesions in real time intraoperatively. HIRI is intimately associated with oxidative stress which impairs lysosomal degradative function, leading to significant changes in lysosomal viscosity. Therefore, lysosomal viscosity is a potential biomarker for the precise targeting of HIRI. Hence, we developed a viscosity-activatable second near-infrared window fluorescent probe (NP-V) for the detection of lysosomal viscosity in hepatocytes and mice during HIRI. A reactive oxygen species–malondialdehyde–cathepsin B signaling pathway during HIRI was established. We further conducted high signal-to-background ratio NIR-II fluorescence imaging of HIRI mice. The contour and boundary of liver lesions were delineated, and as such the precise intraoperative resection of the lesion area was implemented. This research demonstrates the potential of NP-V as a dual-functional probe for the elucidation of HIRI pathogenesis and the direct navigation of HIRI lesions in clinical applications.

Choroidal Disorders in Staphyloma and Dome-Shaped Macula

Posterior staphyloma is described as an ectasia of the limited area of the eyewall mostly in highly myopic eyes. Dome-shaped macula (DSM), detected by optical coherence tomography (OCT) is a convex elevation of the eyewall mostly in the posterior staphyloma area. Both posterior staphyloma and DSM are also rarely seen in eyes without high myopia. The process that begins with the axial enlargement of the Bruch’s membrane (BM) in the equatorial region during axial elongation and breaks in BM play a role in the pathogenesis. It is thought that DSM may be a kind of compensation mechanism in eyes with high myopia and posterior staphyloma. The choroidal layer is relatively thinner in staphyloma and thicker in KŞM. Multimodal imaging is may be required for diagnosis. In OCT, it is important to evaluate the vertical and horizontal scans together and not to keep the scanning area within the lesion boundaries. Attention should be paid to the changes in the posterior eye wall curvature and choroidal thickness. The gradual decrease of choroidal thickness from the periphery to the staphyloma border is an important tip for diagnosis. The fact that the increase in choroidal thickness is limited to the convex elevation area in DSM with serous macular detachment should be considered in differential diagnosis with central serous chorioretinopathy. Studies on increasing the biomechanical power of the sclera in staphyloma are ongoing. It may be a good option to avoid overtreatment in the presence of complications, considering the benefit-risk ratio.

A Transformer Network Architecture for Dermoscopy Image Segmentation

Aiming at the problems of irregular shape and blurred boundary of skin lesions in skin lesions images, this paper proposes a skin lesion segmentation algorithm combining CNN and Transformer. Firstly, Resnet is used as the backbone feature extraction network to extract features, and the extracted feature map sequence is used as the input of Transformer. A new structural boundary attention gate is added to Transformer to extract enough local details to deal with fuzzy boundaries. Finally, DenseASPP is used to enhance features Represents and processes multi-scale information, and proposes an improved loss function, the purpose of which is to make the model pay attention to the boundary region when calculating the loss function. The experimental results show that the dice value and JI value of the network on the ISIC2017 dataset are 0.854534 and 0.767901, respectively, and the dice value and JI value on the ISIC2018 dataset are 0.908548 and 0.843689, respectively, which achieves good results compared to other advanced models. Its effectiveness is proved by comparing with different models and showing the effect.

Differential diagnostic value of periampullary mass: A nomogram established by random forest based on clinical characteristics and contrast‐enhanced ultrasound

To develop a nomogram model for distinguishing benign from malignant ampullary lesions more intuitive and accurate. A total of 124 patients with periampullary lesions from January 2016 to June 2020 were enrolled in this retrospective study. Their clinical information, ultrasound (US), dual contrast-enhanced ultrasound (DCEUS) and MRI image features were used for research. Twenty features were collected in our study. Random forest was used to select the first five most important indicators to construct the prediction model. Patients' age, common bile duct (CBD) diameter, the shape, vascularity, and boundary of lesion, lesion size with or without enlarged after CEUS, the enhancement patterns of arterial phase, the washout patterns of venous phase, CEUS diagnosis, and MRI diagnosis were statistically significant (p < 0.05). After screening for statistically significant indicators by random forest, the first five most important indicators were age, CBD diameter, the enhancement patterns of arterial phase, the washout patterns of venous phase, lesion size with or without enlarged after CEUS, which were used to construct nomogram. The area under curves (AUC) and 95% confidence intervals (CI) for nomogram, MRI + MRCP + DCEUS, DCEUS, MRI + MRCP were 0.98(0.94-1.00), 0.91(0.84-0.97), 0.89(0.80-0.98), 0.68(0.60-0.77), respectively. The sensitivity and specificity were 100.00% and 84.62% for nomogram, 88.29% and 92.31% for MRI + MRCP+DCEUS, 86.49% and 92.31% for DCEUS, 51.35%, and 100.00% for MRI + MRCP. We combined clinical indicators, gray-scale ultrasound characteristics, and CEUS characteristics to build the nomogram, which can be intuitively and accurately used for preoperative malignant prediction of ampullary lesion patients, worthy of clinical generalizability and application.

Active contour model of breast cancer DCE‐MRI segmentation with an extreme learning machine and a fuzzy C‐means cluster

Due to the low contrast, blurred boundary and intensity inhomogeneity of the images, accurate segmentation of breast cancer lesions with dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) still has great challenges. This paper proposed an improved active contour model (ACM) for segmenting breast cancer lesions in DCE-MRI images. First, based on the extreme learning machine (ELM) method, a robust function is proposed that combines image intensities and time-domain features to enhance the difference between the lesions and other tissues. Second, an edge-stop function (ESF) is introduced by combining the image intensity, time-domain feature, and Hessian shape index to detect the irregular and blurred boundaries. At the boundary of breast cancer lesions, the energy function of ACM is minimized and the evolution of the contour curve completes, so the accurate lesion region of breast cancer can be segmented. The mean Dice similar coefficient (DICE), Jaccard similarity (JC) and Hausdorff distance (HD) of the segmentation of the proposed model in 50 samples are 85.88±6.62%, 75.72±9.68% and 11.62±4.72 mm, respectively. The results segmented by the proposed ACM are more similar to the manual segmentation than the compared models.

Skin Lesion Segmentation Based on Edge Attention Vnet with Balanced Focal Tversky Loss

Segmentation of skin lesions from dermoscopic images plays an essential role in the early detection of skin cancer. However, skin lesion segmentation is still challenging due to artifacts such as indistinguishability between skin lesion and normal skin, hair on the skin, and reflections in the obtained dermoscopy images. In this study, an edge attention network (ET-Net) combining edge guidance module (EGM) and weighted aggregation module is added to the 2D volumetric convolutional neural network (Vnet 2D) to maximize the performance of skin lesion segmentation. In addition, the proposed fusion model presents a new fusion loss function by combining balanced binary cross-entropy (BBCE) and focal Tversky loss (FTL). The proposed model has been tested on the ISIC 2018 Task 1 Lesion Boundary Segmentation Challenge dataset. The proposed model outperformed the state-of-the-art studies as a result of the tests.

FECC-Net: A Novel Feature Enhancement and Context Capture Network Based on Brain MRI Images for Lesion Segmentation

In recent years, the increasing incidence of morbidity of brain stroke has made fast and accurate segmentation of lesion areas from brain MRI images important. With the development of deep learning, segmentation methods based on the computer have become a solution to assist clinicians in early diagnosis and treatment planning. Nevertheless, the variety of lesion sizes in brain MRI images and the roughness of the boundary of the lesion pose challenges to the accuracy of the segmentation algorithm. Current mainstream medical segmentation models are not able to solve these challenges due to their insufficient use of image features and context information. This paper proposes a novel feature enhancement and context capture network (FECC-Net), which is mainly composed of an atrous spatial pyramid pooling (ASPP) module and an enhanced encoder. In particular, the ASPP model uses parallel convolution operations with different sampling rates to enrich multi-scale features and fully capture image context information in order to process lesions of different sizes. The enhanced encoder obtains deep semantic features and shallow boundary features in the feature extraction process to achieve image feature enhancement, which is helpful for restoration of the lesion boundaries. We divide the pathological image into three levels according to the number of pixels in the real mask area and evaluate FECC-Net on an open dataset called Anatomical Tracings of Lesions After Stroke (ATLAS). The experimental results show that our FECC-Net outperforms mainstream methods, such as DoubleU-Net and TransUNet. Especially in small target tasks, FECC-Net is 4.09% ahead of DoubleU-Net on the main indicator DSC. Therefore, FECC-Net is encouraging and can be relied upon for brain MRI image applications.

A modified DeepLabV3+ based semantic segmentation of chest computed tomography images for COVID-19 lung infections.

Coronavirus disease (COVID‐19) affects the lives of billions of people worldwide and has destructive impacts on daily life routines, the global economy, and public health. Early diagnosis and quantification of COVID‐19 infection have a vital role in improving treatment outcomes and interrupting transmission. For this purpose, advances in medical imaging techniques like computed tomography (CT) scans offer great potential as an alternative to RT‐PCR assay. CT scans enable a better understanding of infection morphology and tracking of lesion boundaries. Since manual analysis of CT can be extremely tedious and time‐consuming, robust automated image segmentation is necessary for clinical diagnosis and decision support. This paper proposes an efficient segmentation framework based on the modified DeepLabV3+ using lower atrous rates in the Atrous Spatial Pyramid Pooling (ASPP) module. The lower atrous rates make receptive small to capture intricate morphological details. The encoder part of the framework utilizes a pre‐trained residual network based on dilated convolutions for optimum resolution of feature maps. In order to evaluate the robustness of the modified model, a comprehensive comparison with other state‐of‐the‐art segmentation methods was also performed. The experiments were carried out using a fivefold cross‐validation technique on a publicly available database containing 100 single‐slice CT scans from >40 patients with COVID‐19. The modified DeepLabV3+ achieved good segmentation performance using around 43.9 M parameters. The lower atrous rates in the ASPP module improved segmentation performance. After fivefold cross‐validation, the framework achieved an overall Dice similarity coefficient score of 0.881. The results demonstrate that several minor modifications to the DeepLabV3+ pipeline can provide robust solutions for improving segmentation performance and hardware implementation.