Fully Convolutional Network for the Semantic Segmentation of Medical Images: A Survey

AI-powered innovations in pancreatitis imaging: a comprehensive literature synthesis

Early identification of pancreatitis remains a significant clinical diagnostic challenge that impacts patient outcomes. The evolution of quantitative imaging followed by deep learning models has shown great promise in the non-invasive diagnosis of pancreatitis and its complications. We provide an overview of advancements in diagnostic imaging and quantitative imaging methods along with the evolution of artificial intelligence (AI). In this article, we review the current and future states of methodology and limitations of AI in improving clinical support in the context of early detection and management of pancreatitis.

Development and Validation of Deep Learning Preoperative Planning Software for Automatic Lumbosacral Screw Selection Using Computed Tomography

Achieving precise pedicle screw placement in posterior lumbar interbody fusion (PLIF) is essential but difficult due to the intricacies of manual preoperative planning with CT scans. We analyzed CT data from 316 PLIF patients, using Mimics software for manual planning by two surgeons. A deep learning model was trained on 228 patients and validated on 88 patients, assessing planning efficiency and accuracy. Automatic planning successfully segmented and placed screws in all 316 cases, significantly outperforming manual planning in speed. The Dice coefficient for segmentation accuracy was 0.95. The difference in mean pedicle transverse angle (PTA) and pedicle sagittal angle (PSA) for automatic planning screws compared to manual planning screws was 1.63 ± 0.83° and 1.39 ± 1.03°, respectively, and these differences were either statistically comparable or not significantly different compared to the variability of manual planning screws. The average Dice coefficient of implanted screws was 0.63 ± 0.08, and the consistency between automatic screws and manual reference screws was higher than that of internal screws (Dice 0.62 ± 0.09). Compared with manual screws, automatic screws were shorter (46.58 ± 3.09 mm) and thinner (6.24 ± 0.35 mm), and the difference was statistically significant. In qualitative validation, 97.7% of the automatic planning screws were rated Gertzbein-Robbins (GR) Class A and 97.3% of the automatic planning screws were rated Badu Class 0. Deep learning software automates lumbosacral pedicle screw planning, enhancing surgical efficiency and accuracy.

Feature-based detection of breast cancer using convolutional neural network and feature engineering

Breast cancer (BC) is a prominent cause of female mortality on a global scale. Recently, there has been growing interest in utilizing blood and tissue-based biomarkers to detect and diagnose BC, as this method offers a non-invasive approach. To improve the classification and prediction of BC using large biomarker datasets, several machine-learning techniques have been proposed. In this paper, we present a multi-stage approach that consists of computing new features and then sorting them into an input image for the ResNet50 neural network. The method involves transforming the original values into normalized values based on their membership in the Gaussian distribution of healthy and BC samples of each feature. To test the effectiveness of our proposed approach, we employed the Coimbra and Wisconsin datasets. The results demonstrate efficient performance improvement, with an accuracy of 100% and 100% using the Coimbra and Wisconsin datasets, respectively. Furthermore, the comparison with existing literature validates the reliability and effectiveness of our methodology, where the normalized value can reduce the misclassified samples of ML techniques because of its generality.

Semantic segmentation in crystal growth process using fake micrograph machine learning

Microscopic evaluation is one of the most effective methods in materials research. High-quality images are essential to analyze microscopic images using artificial intelligence. To overcome this challenge, we propose the machine learning of "fake micrographs" in this study. To verify the effectiveness of this method, we chose to analyze the optical microscopic images of the crystal growth process of a Ge thin film, which is a material in which it is difficult to obtain a contrast between the crystal and amorphous states. By learning the automatically generated fake micrographs that mimic the crystal growth process, the machine learning model can now identify the low-resolution real micrographs as crystalline or amorphous. Comparing the three types of machine learning models, it was found that ResUNet ++ exhibited high accuracy, exceeding 90%. The technology developed in this study for the automatic and rapid analysis of low-resolution images is widely helpful in material research.

Parotid Gland Segmentation Using Purely Transformer-Based U-Shaped Network and Multimodal MRI

Parotid gland tumors account for approximately 2% to 10% of head and neck tumors. Segmentation of parotid glands and tumors on magnetic resonance images is essential in accurately diagnosing and selecting appropriate surgical plans. However, segmentation of parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. Recently, deep learning has developed rapidly, and Transformer-based networks have performed well on many computer vision tasks. However, Transformer-based networks have yet to be well used in parotid gland segmentation tasks. We collected a multi-center multimodal parotid gland MRI dataset and implemented parotid gland segmentation using a purely Transformer-based U-shaped segmentation network. We used both absolute and relative positional encoding to improve parotid gland segmentation and achieved multimodal information fusion without increasing the network computation. In addition, our novel training approach reduces the clinician's labeling workload by nearly half. Our method achieved good segmentation of both parotid glands and tumors. On the test set, our model achieved a Dice-Similarity Coefficient of 86.99%, Pixel Accuracy of 99.19%, Mean Intersection over Union of 81.79%, and Hausdorff Distance of 3.87. The purely Transformer-based U-shaped segmentation network we used outperforms other convolutional neural networks. In addition, our method can effectively fuse the information from multi-center multimodal MRI dataset, thus improving the parotid gland segmentation.

Deep learning for MRI lesion segmentation in rectal cancer

Rectal cancer (RC) is a globally prevalent malignant tumor, presenting significant challenges in its management and treatment. Currently, magnetic resonance imaging (MRI) offers superior soft tissue contrast and radiation-free effects for RC patients, making it the most widely used and effective detection method. In early screening, radiologists rely on patients' medical radiology characteristics and their extensive clinical experience for diagnosis. However, diagnostic accuracy may be hindered by factors such as limited expertise, visual fatigue, and image clarity issues, resulting in misdiagnosis or missed diagnosis. Moreover, the distribution of surrounding organs in RC is extensive with some organs having similar shapes to the tumor but unclear boundaries; these complexities greatly impede doctors' ability to diagnose RC accurately. With recent advancements in artificial intelligence, machine learning techniques like deep learning (DL) have demonstrated immense potential and broad prospects in medical image analysis. The emergence of this approach has significantly enhanced research capabilities in medical image classification, detection, and segmentation fields with particular emphasis on medical image segmentation. This review aims to discuss the developmental process of DL segmentation algorithms along with their application progress in lesion segmentation from MRI images of RC to provide theoretical guidance and support for further advancements in this field.

Construction of deep learning-based convolutional neural network model for automatic detection of fluid hysteroscopic endometrial micropolyps in infertile women with chronic endometritis

OBJECTIVE(S)

Chronic endometritis (CE) is a localized mucosal inflammatory disorder associated with female infertility of unknown etiology, endometriosis, tubal factors, repeated implantation failure, and recurrent pregnancy loss, along with atypical uterine bleeding and iron deficiency anemia. Diagnosis of CE has traditionally relied on endometrial biopsy and detection of CD138(+) endometrial stromal plasmacytes. To develop a less invasive diagnostic system for CE, we aimed to construct a deep learning-based convolutional neural network (CNN) model for the automatic detection of endometrial micropolyps (EMiP), a fluid hysteroscopy (F-HSC) finding recognized as tiny protrusive lesions that are closely related to this disease.

STUDY DESIGN

This is an in silico study using archival images of F-HSC performed at an infertility center in a private clinic. A total of 244 infertile women undergoing F-HSC on the days 6-12 of the menstrual cycle between April 2019 and December 2021 with histopathologically-confirmed CE with the aid of immunohistochemistry for CD138 were utilized.

RESULTS

The archival F-HSC images of 208 women (78 with EMiP and 130 without EMiP) who met the inclusion criteria were finally subjected to analysis. Following preprocessing of the images, half a set was input into a CNN architecture for training, whereas the remaining images were utilized as the test set to evaluate the performance of the model, which was compared with that of the experienced gynecologists. The sensitivity, specificity, accuracy, precision, and F1-score of the CNN model-aided diagnosis were 93.6 %, 92.3 %, 92.8 %, 88.0 %, and 0.907, respectively. The area under the receiver operating characteristic curves of the CNN model-aided diagnosis (0.930) was at a similar level (p > .05) to the value of conventional diagnosis by three experienced gynecologists (0.927, 0.948, and 0.906).

CONCLUSION

These findings indicate that our deep learning-based CNN is capable of recognizing EMiP in F-HSC images and holds promise for further development of the computer-aided diagnostic system for CE.

Identifying Lymph Nodes and Their Statuses from Pretreatment Computer Tomography Images of Patients with Head and Neck Cancer Using a Clinical-Data-Driven Deep Learning Algorithm

BACKGROUND

Head and neck cancer is highly prevalent in Taiwan. Its treatment mainly relies on clinical staging, usually diagnosed from images. A major part of the diagnosis is whether lymph nodes are involved in the tumor. We present an algorithm for analyzing clinical images that integrates a deep learning model with image processing and attempt to analyze the features it uses to classify lymph nodes.

METHODS

We retrospectively collected pretreatment computed tomography images and surgery pathological reports for 271 patients diagnosed with, and subsequently treated for, naïve oral cavity, oropharynx, hypopharynx, and larynx cancer between 2008 and 2018. We chose a 3D UNet model trained for semantic segmentation, which was evaluated for inference in a test dataset of 29 patients.

RESULTS

We annotated 2527 lymph nodes. The detection rate of all lymph nodes was 80%, and Dice score was 0.71. The model has a better detection rate at larger lymph nodes. For those identified lymph nodes, we found a trend where the shorter the short axis, the more negative the lymph nodes. This is consistent with clinical observations.

CONCLUSIONS

The model showed a convincible lymph node detection on clinical images. We will evaluate and further improve the model in collaboration with clinical physicians.

Development and Validation of an Artificial Intelligence Preoperative Planning and Patient-Specific Instrumentation System for Total Knee Arthroplasty

BACKGROUND

Accurate preoperative planning for total knee arthroplasty (TKA) is crucial. Computed tomography (CT)-based preoperative planning offers more comprehensive information and can also be used to design patient-specific instrumentation (PSI), but it requires well-reconstructed and segmented images, and the process is complex and time-consuming. This study aimed to develop an artificial intelligence (AI) preoperative planning and PSI system for TKA and to validate its time savings and accuracy in clinical applications.

METHODS

The 3D-UNet and modified HRNet neural network structures were used to develop the AI preoperative planning and PSI system (AIJOINT). Forty-two patients who were scheduled for TKA underwent both AI and manual CT processing and planning for component sizing, 20 of whom had their PSIs designed and applied intraoperatively. The time consumed and the size and orientation of the postoperative component were recorded.

RESULTS

The Dice similarity coefficient (DSC) and loss function indicated excellent performance of the neural network structure in CT image segmentation. AIJOINT was faster than conventional methods for CT segmentation (3.74 ± 0.82 vs. 128.88 ± 17.31 min, p < 0.05) and PSI design (35.10 ± 3.98 vs. 159.52 ± 17.14 min, p < 0.05) without increasing the time for size planning. The accuracy of AIJOINT in planning the size of both femoral and tibial components was 92.9%, while the accuracy of the conventional method in planning the size of the femoral and tibial components was 42.9% and 47.6%, respectively (p < 0.05). In addition, AI-based PSI improved the accuracy of the hip-knee-ankle angle and reduced postoperative blood loss (p < 0.05).

CONCLUSION

AIJOINT significantly reduces the time needed for CT processing and PSI design without increasing the time for size planning, accurately predicts the component size, and improves the accuracy of lower limb alignment in TKA patients, providing a meaningful supplement to the application of AI in orthopaedics.

[Application of U-Net network in automatic image segmentation of adenoid and airway of nasopharynx]

Objective:To explore the effect of fully automatic image segmentation of adenoid and nasopharyngeal airway by deep learning model based on U-Net network. Methods:From March 2021 to March 2022, 240 children underwent cone beam computed tomography（CBCT） in the Department of Otolaryngology, Head and Neck Surgery, General Hospital of Shenzhen University. 52 of them were selected for manual labeling of nasopharynx airway and adenoid, and then were trained and verified by the deep learning model. After applying the model to the remaining data, compare the differences between conventional two-dimensional indicators and deep learning three-dimensional indicators in 240 datasets. Results:For the 52 cases of modeling and training data sets, there was no significant difference between the prediction results of deep learning and the manual labeling results of doctors（P>0.05）. The model evaluation index of nasopharyngeal airway volume: Mean Intersection over Union（MIOU） s （86.32±0.54）%; Dice Similarity Coefficient（DSC）: （92.91±0.23）%; Accuracy: （95.92±0.25）%; Precision: （91.93±0.14）%; and the model evaluation index of Adenoid volume: MIOU: （86.28±0.61）%; DSC: （92.88±0.17）%; Accuracy: （95.90±0.29）%; Precision: （92.30±0.23）%. There was a positive correlation between the two-dimensional index A/N and the three-dimensional index AV/（AV+NAV） in 240 children of different age groups（P<0.05）, and the correlation coefficient of 9-13 years old was 0.74. Conclusion:The deep learning model based on U-Net network has a good effect on the automatic image segmentation of adenoid and nasopharynx airway, and has high application value. The model has a certain generalization ability.