Current Medical Imaging - Volume 19, Issue 12, 2023

Machine Learning (ML) plays an essential part in the research area of medical image processing. The advantages of ML techniques lead to more intelligent, accurate, and automatic computeraided detection (CAD) systems with improved learning capability. In recent years, deep learning-based ML approaches developed to improve the diagnostic capabilities of CAD systems. This study reviews image enhancement, ML and DL methods for breast cancer detection and diagnosis using mammogram images and provides an overview of these methods. The analysis of different ways of ML and DL shows that the usages of traditional ML approaches are limited. However, DL techniques have an excellent future for implementing medical image analysis and improving the ability to exist CAD systems. Despite the significant advancements in deep learning methods for analyzing medical images to detect breast cancer, challenges still exist regarding data quality, computational cost, and prediction accuracy.

Background: Many studies have shown that vertebral trabecular attenuation measured on CT scan corresponds well to DXA results for bone mineral density. These studies were based on crosssectional data. Hence, there were limitations in explaining the constantly changing vertebral trabecular attenuation from CT and T-score from DXA over time. Objective: This study aimed to determine the longitudinal association between the vertebral trabecular attenuation measured on computed tomography (CT) and the T-score measured by dual-energy X-ray absorptiometry (DXA). Methods: We performed a database search for 333 patients who underwent surgery for breast cancer, preoperative treatment, and at least one follow-up chest CT and DXA from January, 2013 through May, 2021. One musculoskeletal radiologist measured the mean vertebral trabecular attenuation of lumbar vertebra 1(L1) on axial unenhanced images at the pedicle level by manually placing the region of interest (ROI). DXA of the lumbar spine was performed, and the lowest T-score of the lumbar spine was used for the analysis. We evaluated the association between L1 trabecular attenuation from chest CT and T-score from DXA over time using the generalized estimating equations (GEE) model to analyze longitudinal corrected data. Results: A total of 150 women (mean age, 52.4 ± 11.0 years) were included. There was a statistically significant association between L1 trabecular attenuation from chest CT and T-score from DXA in the unadjusted model (p < 0.001) and adjusted model (p < 0.001). T-score value increased by 0.172 (95% confidence interval (CI): 0.145-0.200, p < 0.001) per 10 unit (HU) of L1 trabecular attenuation at time = 0 in unadjusted model and by 0.173 (95% CI: 0.143-0.203, p < 0.001) in all adjusted model. Conclusion: We demonstrated that L1 attenuation from chest CT images was longitudinally associated with T-score from DXA, and the degree of association appeared to be decreased over time in breast cancer patients regardless of their medical condition.

Background: As the largest concentration of neural stem cells in adult brain, the subventricular zone (SVZ) is considered to be a potential source of glioblastoma (GBM) occurrence in recent years. Methods: In this study, 116 patients with glioblastoma treated at PLA General Hospital were retrospectively reviewed. The features of SVZ contacting glioblastoma were analyzed in terms of MR imaging and MGMT promoter methylation. We also evaluated the prognostic value of SVZ contacting in GBM patients. Results: GBM with SVZ involvement on MRI is more likely to grow across the midline (36.8% vs. 6.9%, P=0.002), more often multifocal lesion (35.6% vs. 6.9%, P=0.003) and have a lower proportion of MGMT promoter methylation (36.8% vs. 69.0%, P=0.003). The median overall survival and progression- free survival of patients in the SVZ contacting group were 12 months and 7 months, while 25 months and 17 months in the non-contacting group (P<0.001, respectively). There was no significant difference in overall survival (P=0.229) and progression-free survival (P=0.808) between patients with different SVZ contacting regions. Multivariate survival analysis indicated that patients with MRI SVZ involvement showed worse overall survival (HR=2.060, 95%CI 1.195-3.550, P=0.009) and progression- free survival (HR=3.021, 95%CI 1.788-5.104, P < 0.001). Conclusion: This study suggested that MRI SVZ involvement at diagnosis is an independent risk factor for overall survival and progression-free survival in IDH wild-type glioblastoma patients. Based on MR imaging, we also found that SVZ contacting glioblastomas had a larger proportion of crossing midline tumors and multifocal lesions. In addition, patients with SVZ contact in our research presented a lower proportion of MGMT promoter methylation.

Background: Facial information acquired via three-dimensional reconstruction of head computed tomography (CT) data may be considered personal information, which can be problematic for neuroimaging studies. However, no study has verified the relationship between slice thickness and face reproducibility. This study determined the relationship and match rate between image slice thickness and face detection accuracy of face-recognition software in facial reconstructed models. Methods: Head CT data of 60 cases comprising entire faces obtained under conditions of non-contrast and 1-mm slice thickness were resampled to obtain 2-10-mm slice-thickness data. Facial models, reconstructed by image thresholding, were acquired from the data. We performed face detection tests per slice thickness on the models and calculated the face detection rate. The reconstructed facial models created from 1-mm slice-thickness data and other slice thicknesses were used as training and test data, respectively. Match confidence scores were obtained via three programs, match rates were calculated per slice thickness, and generalized estimating equations were used to evaluate the match rate trend. Results: In general, the face detection rates for the 1-10-mm slice thicknesses were 100, 100, 98.3, 98.3, 95.0, 91.7, 86.7, 78.3, 68.3, and 61.7 %, respectively. The match rates for the 2-10-mm slice thicknesses were 100, 98.3, 98.3, 95.0, 85.0, 71.7, 53.3, 28.3, and 16.7 %, respectively. Conclusion: The reconstructed models tended to have higher match rates as the slice thickness decreased. Thus, thin-slice head CT imaging data may increase the possibility of the information becoming personally identifiable health information.

Objective: to investigate the feasibility of gadoxetic acid (Gd-EOB-DTPA) enhanced MRI combined with T1 mapping in quantitative hepatic function assessment. Methods: this study retrospectively enrolled 94 patients with Gd-EOB-DTPA enhanced MRI combined with T1 mapping, divided into group A (grade A, n=73), group B (grade B, n=14) and group C (grade C, n=7) based on Child-Pugh classification. Liver T1 relaxation times on plain scan (T1P) and hepatocellular phase (T1E) were measured. Decrease in T1 (T1D) and the percentage of decrease in T1 (T1D%) were calculated as follows: T1D=T1P-T1E, T1D%= T1D/T1P×100%. The relationship between T1P, T1E, T1D, T1D% and liver function classification was analyzed. Results: T1P, T1D, and T1D% in group A were significantly higher than those of group B and C. T1E in group A was lower than those of group B and C. T1D% was significantly different between group B and C. There was no significant difference in T1P, T1E, T1D between groups B and C. T1E was positively correlated with liver function levels, T1P and T1D had a negative correlation with liver function levels. T1P, T1E, T1D, T1D% were significantly different between cirrhotic and non-cirrhotic groups. T1D% of less than 70% suggests liver dysfunction. Conclusion: Gd-EOB-DTPA enhanced liver MRI combined with T1 mapping is feasible for quantitative assessment of hepatic function.

Objective: To explore the diagnostic value and severity of acute kidney injury (AKI) in patients with acute pancreatitis (AP) using intravoxel incoherent motion imaging (IVIM), diffusion tensor imaging (DTI), and diffusion kurtosis imaging (DKI). Methods: 224 AP patients, categorized into either the AKI group or the non-AKI group, were retrospectively analyzed in this study. MRI sequences included routine abdominal, IVIM, DTI, and DKI scans, and the main MRI parameters of kidney imaging and clinical characteristics were measured. The diagnostic performance of AKI was compared, and the relationships among these indices, glomerular filtration rate (eGFR), and AKI staging were analyzed. Finally, all parameters were analyzed by single and multi-parameter regression. Results: Compared with the non-AKI group, the fast apparent diffusion coefficient (ADC) value and perfusion fraction (Ff ADC) value of the renal medulla in the AKI group were significantly lower than those in the non-AKI group. The fractional anisotropy (FA) value in the renal cortex was significantly lower than that in the medulla and significantly lower than in the non-AKI group. Lastly, the renal medulla mean kurtosis (MK) value was also significantly lower in the AKI group compared to the non- AKI group and exhibited the best diagnostic value for AKI in AP patients. The renal medulla MK value positively correlated with AKI staging and negatively correlated with eGFR. The MK value was an independent risk factor for AKI, as evidenced by multi-parameter logistic regression analysis. Conclusion: The measurement of renal DKI parameters is practical for diagnosing and predicting the severity of acute kidney injury in AP patients.

Background: Diffuse leptomeningeal glioneuronal tumors (DL-GNT) are rare glioneuronal neoplasms with oligodendroglioma-like cells. These tumors can present as a dominant intracranial mass or as a solitary spinal cord mass without leptomeningeal involvement. In this study, we aimed to determine the magnetic resonance imaging and histopathological features, treatment modalities, and clinical outcomes of the parenchymal forms of DL-GNTs. Methods: This is a retrospective three-center case series study of 5 patients with a confirmed parenchymal form of DLGTs, out of which 4 patients were adults. Brain and spinal cord MR imaging were performed in all patients at either 1.5 or 3T. The patients' age ranged from 5 years to 50 years with a mean age of 27.6 years at presentation. Results: Four of the tumors were located in the frontal lobe, and one in the tectum. They were usually solid-cystic enhancing tumors as the other mixed neuronal-glial tumors. All of the tumors had an extension to the superficial surface of a cerebral hemisphere. One had systemic bone metastases. The clinical signs and symptoms of the parenchymal form varied based on the location of the mass, in contrast to the leptomeningeal form associated with hydrocephalus. In one case, the tumor’s initial grade was defined as intermediate. The initial histopathology of the two cases was low-grade and no upgrade occurred in the follow-up period. In two cases, although the tumors were low grade initially, they progressed to an anaplastic form in the follow-up period. Conclusion: The parenchymal form of DL-GNTs is common in adults. Extension to the superficial surface of a cerebral hemisphere is a distinctive imaging feature. Systemic osseous metastasis may occur. Due to the presence of common histopathological features, including the biphasic composition of glial and neuronal cell elements and oligodendroglioma-like cells, a proposed classification approach might be more beneficial for the histopathological and imaging description, and management of the glioneuronal tumors with oligodendroglioma-like features.

Background: PET imaging is one of the most widely used neurological disease screening and diagnosis techniques. Aims: Since PET involves the radiation and tolerance of different people, the improvement that has always been focused on is to cut down radiation, in the meantime, ensuring that the generated images with low-dose tracer and generated images with standard-dose tracer have the same details of images. Methods: We propose a lightweight low-dose PET super-resolution network (SRPET-Net) based on a convolutional neural network. In this research, We propose a method for accurately recovering highresolution (HR) PET images from low-resolution (LR) PET images. The network learns the details and structure of the image between low-dose PET images and standard-dose PET images and, afterward, reconstructs the PET image by the trained network model. Results: The experiments indicate that the SRPET-Net can achieve a superior peak signal-to-noise ratio (PSNR) and structural similarity index measurement (SSIM) values. Moreover, our method has less memory consumption and lower computational cost. Conclusion: In our follow-up work, the technology can be applied to medical imaging in many different directions.

Background: Traditional radiography angles do not adequately reveal the shape and position of the right ventricular pacing electrode. Objective: This study aimed to explore better radiography angles with the help of cardiac computed tomography (CT). Methods: We analyzed the cardiac CT images of consecutive outpatients from 2018 to 2019. The right anterior oblique (RAO) 30° and the left anterior oblique (LAO) 40° were found to sufficiently display the shape and position of the right ventricular pacing electrode. Results: A total of 214 consecutive outpatients were enrolled, whose average age was 55.0 ± 13.0 years, and 151 were male (70.6%). Through analyzing the cardiac CT images, the α angle (33.7° ± 6.1) and the γ angle (38.8° ± 8.0) were determined. Furthermore, we verified these angles in 48 patients after pacemaker implantation. The results showed that the ratio of the length of right ventricular electrode using the RAO α angle (≈30°) to the posterior-anterior position (PA position) was 1.099 ± 0.157 vs. 1.053 ± 0.182 (the ratio using the traditional RAO 45°) (P < 0.001). We observed that the relationship between the right ventricular active electrode and the ventricular septum was better identified using the LAO γ angle (≈40°) than the traditional 60° angle. Conclusion: With the help of cardiac CT, we found that RAO 30° could better show the shape and length of the right ventricular pacing electrode, and LAO 40° could better show the positional relationship between the pacing electrode and the ventricular septum.

Background: Point shear wave elastography (pSWE) is used to provide quantitative information about the elasticity of the tissues. It has been used in many clinical applications to identify diseases early. This study aims to assess the suitability of pSWE to evaluate pancreas tissue stiffness and provide normal reference values for healthy pancreas tissues. Methods: This study was conducted in the diagnostic department at a tertiary care hospital between October and December 2021. Sixteen healthy volunteers (eight men and eight women) participated. Pancreas elasticity measurements were obtained in different regions of interest (head, body, and tail). Scanning was performed by a certified sonographer using Philips EPIC7 ultrasound systems (Philips Ultrasound; Bothel, WA, USA). Results: The mean velocity of the head of the pancreas was 1.3 ± 0.3 m/s (median: 1.2 m/s), the body was 1.4 ± 0.3 m/s (median: 1.4 m/s), and the tail was 1.4 ± 0.4 m/s (median: 1.2 m/s). The mean dimension was 17 ± 3 mm, 14 ± 4 mm, and 14 ± 6 mm for the head, body, and tail. The correlation of pancreas velocity across different segments and dimensions shows no significant difference with p values of 0.39 and 0.11, respectively. Conclusion: This study shows that assessing the elasticity of the pancreas using pSWE is possible. A combination of SWVs measurements and dimensions could provide an early evaluation of pancreas status. Further studies including pancreatic disease patients are recommended.

Purpose: This study proposes a method for improving the accuracy of three-dimensional (3D) models generated through cone-beam computed tomography (CBCT). Methods: A 3D cuboid model fitted with a ¼-scale dentition on its top surface was constructed to simulate an alveolar bone with teeth. A physical specimen of the model was printed and the distance between its opposite sides was measured using a vernier caliper. The physical model was light-scanned, and the surface data of the generated 3D model were corrected by calibrating the distance between opposite sides against the vernier caliper measurements. The physical model was also scanned using CBCT to reconstruct a second 3D model. The overall deviation between the two models and the distance deviation in each direction of the cuboid and dentition were quantified and statistically analyzed. Results: The overall deviation between the reconstructed CBCT model and the calibrated structured light-scanned model was 0.098 ± 0.001 mm. Following calibration, the overall deviation was 0.010 ± 0.006 mm. A one-way variance analysis suggested that the overall deviations' differences were not statistically significant (P < 0.05). Conclusion: This study lays a solid foundation for accurate dental implantation.

Background: Video capsule endoscopy (VCE) is an attractive method for diagnosing and objectively monitoring disease activity in celiac disease (CeD). Its use, facilitated by artificial intelligence- based tools, may allow computer-assisted interpretation of VCE studies, transforming a subjective test into a quantitative and reproducible measurement tool. Objective: To evaluate and compare objective CeD severity assessment as determined with VCE by expert human readers and a machine learning algorithm (MLA). Methods: Patients ≥ 18 years with histologically proven CeD underwent VCE. Examination frames were scored by three readers from one center and the MLA, using a 4-point ordinal scale for assessing the severity of CeD enteropathy. After scoring, curves representing CeD severity across the entire small intestine (SI) and individual tertiles (proximal, mid, and distal) were fitted for each reader and the MLA. All comparisons used Krippendorff’s alpha; values > 0.8 represent excellent to ‘almost perfect’ inter-reader agreement. Results: VCEs from 63 patients were scored. Readers demonstrated strong inter-reader agreement on celiac villous damage (alpha=0.924), and mean value reader curves showed similarly excellent agreement with MLA curves (alpha=0.935). Average reader and MLA curves were comparable for mean and maximum values for the first SI tertile (alphas=0.932 and 0.867, respectively) and the mean value over the entire SI (alpha=0.945). Conclusion: A novel MLA demonstrated excellent agreement on whole SI imaging with three expert gastroenterologists. An ordinal scale permitted high inter-reader agreement, accurately and reliably replicated by the MLA. Interpreting VCEs using MLAs may allow automated diagnosis and disease burden assessment in CeD.

Objective: Thymic cavernous hemangioma (CH) as a benign lesion is extremely rare, and it is very difficult to distinguish it from other anterior mediastinal tumors on imaging. Case Presentation: We treated a 63-year-old woman with thymic CH that was incidentally found by chest computed tomography (CT). Contrast-enhanced chest CT described an anterior superior mediastinal mass, approximately 2.5cm in diameter, with centripetal enhancement and an increased range of enhancement in the delayed period. The tumor and thymus were completely resected by video-assisted thoracic surgery (VATS) to provide a definitive diagnosis and precise treatment. Postoperative pathology suggested that the tumor was a thymus cavernous hemangioma. Conclusion: We reported an extremely uncommon case of CH in the thymus and reviewed the previously published literature. CH can occur at any age and body tissue, and there is no significant difference between males and females. Most patients are asymptomatic, often found by chest CT. Enhanced chest CT suggested that an increased area of venous enhancement at a prolonged duration may be a feature of CH. Complete surgical resection is a safe and effective method for CH in the thymus.

Introduction/Background: Central venous catheters, which provide a unique convenience in the management of critical patients, have many advantages, as well as early and late complications. Early complications include pneumothorax, vascular perforation, hematoma formation, air embolism, or catheter malposition that may occur during or shortly after catheter insertion. Late complications include infection, venous stenosis, catheter thrombosis, and catheter tip migration. In the literature, embolization of a calcified fibrin sheath due to a central venous catheter to the pulmonary artery has been reported only in one case. Case Presentation: The purpose of this report is to present bilateral pulmonary artery embolism in a patient who presented with cough and chest pain caused by calcified fibrin sheath of the port catheter removed before that was used for regular chemotherapeutic infusions due to liposarcoma, as the second case in the literature with imaging findings. The patient underwent medical treatment, and as a result of the treatment, symptoms regressed. Conclusion: Central venous catheters have many complications, and although it is rare, pulmonary embolism is one of them. The embolism of a calcified sheath is even rare, but it is still possible. However, an embolism can cause significant morbidity and even mortality for a patient if it occurs. As physicians, we should be aware of this entity to diagnose.

Background: I-131 uptake in the breast could be occasionally observed in differentiated thyroid cancer who received I-131 scintigraphy or therapy. Herein, we report a postpartum patient with papillary thyroid cancer with breast uptake who received I-131 therapy. Case Report: A 33-year-old postpartum woman with thyroid cancer received 120mCi(4440MBq) I- 131 therapy 5 weeks after she ceased to breastfeed. The second day after ingestion of I-131, wholebody scintigraphy demonstrated asymmetric and marked uptake in bilateral breasts. The radiation dose of I-131 in the lactating breast would decrease rapidly by reducing the activity of breasts and expressing breast milk with an electric pump once daily. Results: The sixth day after administration, scintigraphy demonstrated weak uptake in bilateral breasts. Conclusion: Physiologic I-131 uptake in the breast could occur in a postpartum woman with thyroid cancer who received I-131 therapy. In this patient, we find that the radiation dose of I-131 accumulated in the lactating breast would decrease rapidly through reducing the activity of breasts and expressing breast milk with an electric pump, which may be a better choice for the postpartum patient who was not given lactation-inhibiting medications and had I-131 therapy.

Purpose: To identify MRI findings for injuries to periarticular soft tissue structures that were related to isolated greater tubercle fracture. Materials and Methods: 16 patients (mean age: 53.8, range 30-71 yrs) were enrolled and diagnosed with isolated greater tubercle (GT) fracture with CT and MRI and underwent shoulder arthroscopy from September 2009 to April 2019. Two musculoskeletal radiologists were blinded to the patient history and arthroscopic surgical findings and reviewed patient’s CT and MRI. Fracture displacement, fracture center, and presence of bony Bankart lesion with Hill sachs lesion were checked on shoulder CT. Soft tissue injuries, including rotator cuff injury, deltoid muscle injury, long head of biceps tendon injury, capsular injury, glenoid injury, and injury location of the subscapularis and infraspinatus were checked on shoulder MRI. Results: MRI showed supraspinatus injury (56.3%), subscapularis injury (56.3%), deltoid muscle injury (25%), infraspinatus injury (25%), teres minor muscle injury (37.5%), injury of the long head of the biceps tendon (43.8%), inferior glenohumeral ligament tear (87.5%), superior labral anterior-toposterior lesion (25%), Bankart lesion (18.8%), and Hill-Sachs lesion (6.3%). 88.9% of subscapularis injuries and 75% of infraspinatus injuries showed caudal predominance. All of the patients with infraspinatus injuries showed concomitant teres minor muscle injuries. In order of frequency, the fracture centers were anterior (25%), posterior (31.3%), and all (43.8%). For patients with Bankart and Hill- Sachs lesions, the fracture center included the posterior portion in all cases. Conclusion: MRI and active arthroscopic examinations may be valuable when an isolated GT avulsion fracture is identified on X-ray examination.