Evaluation of an artificial intelligence U-net algorithm for pulmonary nodule tracking on chest computed tomography images.

OBJECTIVES: To apply image registration in the follow up of lung nodules and verify the feasibility of automatic tracking of lung nodules using an artificial intelligence (AI) method. METHODS: For this retrospective, observational study, patients with pulmonary nodules 5-30 mm in diameter on computed tomography (CT) and who had at least six months follow-up were identified. Two radiologists defined a 'correct' cuboid circumscribing each nodule which was used to judge the success/failure of nodule tracking. An AI algorithm was applied in which a U-net type neural network model was trained to predict the deformation vector field between two examinations. When the estimated position was within a defined cuboid, the AI algorithm was judged a success. RESULTS: In total, 49 lung nodules in 40 patients, with a total of 368 follow-up CT examinations were examined. The success rate for each time evaluation was 94% (345/368) and for 'nodule-by-nodule evaluation' was 78% (38/49). Reasons for a decrease in success rate were related to small nodules and those that decreased in size. CONCLUSION: Automatic tracking of lung nodules is highly feasible.

Magnetic resonance bone imaging: applications to vertebral lesions.

MR bone imaging is a recently introduced technique, that allows visualization of bony structures in good contrast against adjacent structures, like CT. Although CT has long been considered the modality of choice for bone imaging, MR bone imaging allows visualization of the bone without radiation exposure while simultaneously allowing conventional MR images to be obtained. Accordingly, MR bone imaging is expected as a new imaging technique for the diagnosis of miscellaneous spinal diseases. This review presents several sequences used in MR bone imaging including black bone imaging, ultrashort/zero echo time (UTE/ZTE) sequences, and T1-weighted 3D gradient-echo sequence. We also illustrate clinical cases in which spinal lesions could be effectively demonstrated on MR bone imaging, performed in most cases using a 3D gradient-echo sequence at our institution. The lesions presented herein include degenerative diseases, tumors and similar diseases, fractures, infectious diseases, and hemangioma. Finally, we discuss the differences between MR bone imaging and previously reported techniques, and the limitations and future perspectives of MR bone imaging.

Development and performance evaluation of a deep learning lung nodule detection system.

BACKGROUND: Lung cancer is the leading cause of cancer-related deaths throughout the world. Chest computed tomography (CT) is now widely used in the screening and diagnosis of lung cancer due to its effectiveness. Radiologists must identify each small nodule shadow from 3D volume images, which is very burdensome and often results in missed nodules. To address these challenges, we developed a computer-aided detection (CAD) system that automatically detects lung nodules in CT images. METHODS: A total of 1997 chest CT scans were collected for algorithm development. The algorithm was designed using deep learning technology. In addition to evaluating detection performance on various public datasets, its robustness to changes in radiation dose was assessed by a phantom study. To investigate the clinical usefulness of the CAD system, a reader study was conducted with 10 doctors, including inexperienced and expert readers. This study investigated whether the use of the CAD as a second reader could prevent nodular lesions in lungs that require follow-up examinations from being overlooked. Analysis was performed using the Jackknife Free-Response Receiver-Operating Characteristic (JAFROC). RESULTS: The CAD system achieved sensitivity of 0.98/0.96 at 3.1/7.25 false positives per case on two public datasets. Sensitivity did not change within the range of practical doses for a study using a phantom. A second reader study showed that the use of this system significantly improved the detection ability of nodules that could be picked up clinically (p = 0.026). CONCLUSIONS: We developed a deep learning-based CAD system that is robust to imaging conditions. Using this system as a second reader increased detection performance.

MR digital subtraction angiography with three-dimensional data acquisition in the diagnosis of brain tumors: preliminary experience.

MR digital subtraction angiography (DSA) visualizes intracranial vasculature using a rapid T1-weighted sequence and a bolus injection of gadolinium. Although two-dimensional sequences are most frequently used, we applied a three-dimensional technique in combination with a fast method of k-space filling to improve both the temporal and spatial resolutions. In this preliminary study, we assessed the feasibility of using this technique for the diagnosis of brain tumors in 21 patients by reviewing the obtained images and, in 10 patients, comparing the images with conventional angiograms. MR DSA visualized a tumor stain in 11 patients, a tumor mass effect in 9, and tumor-related vessels in 5. In 9 of the 10 patients for whom conventional angiograms were available, the two kinds of angiograms corresponded well. Three-dimensional MR DSA using our technique is a useful adjunct to conventional MR imaging for the visualization of tumor hemodynamics and, in some cases, tumor-related vessels and mass effects.

Application of multi-detector row helical scanning to postmyelographic CT.

The advent of a multi-detector row helical CT has made it possible to attain images over a broader area with good spatial resolution. We assessed whether postmyelographic CT scans obtained using this system provided more information than conventional imaging techniques. Postmyelographic CT scans were preoperatively obtained in 46 patients using a multi-detector row helical CT system. Reconstructed images in the sagittal and coronal planes in all patients and curved reformation images along the dural sac in 37 patients were compared with myelograms and conventional postmyelographic CT scans. In 34 patients comparison was also made with MR images. The multi-detector row CT images demonstrated deformities of the dural sac more clearly than the other modalities in 39 of the 46 patients. They also provided the best visualization of nerve root abnormalities in 24 of the 46 patients and clearly revealed the presence of spurs in all 22 patients with spinal canal stenosis. Postmyelographic CT scans made using a multi-detector row helical CT system provide more information on the dural sac, nerve sleeves, and their contents than other imaging techniques.

MR digital subtraction angiography in the diagnosis of meningiomas.

OBJECTIVE: MR digital subtraction angiography (DSA) is a technique for demonstrating the vasculature combining a rapid two-dimensional T1-weighted sequence with a bolus injection of gadolinium. We attempted to determine its contribution to the diagnosis of intracranial meningiomas. METHODS AND PATIENTS: MR DSA was performed in 18 patients with meningioma as well as in 28 patients with other tumors. The findings were analyzed regarding demonstration of tumor stain and tumor-related vessels. RESULTS: All meningiomas except one were visualized as a homogeneous and intense stain. Feeding arteries were visualized in 2 patients, and draining or abnormal veins in three. In 21 of the 28 patients with other tumors, tumor stains of varying degrees were demonstrated. CONCLUSION: MR DSA can serve as an adjunct to routine MR imaging, because it enables assessment of the hemodynamics of meningiomas and facilitates its differential diagnosis from other tumors.

Visualization of extracranial-intracranial bypass using multidetector-row helical computed tomography angiography.

PURPOSE: The advent of multidetector-row computed tomography (MDCT) has enabled images with good spatial resolution to be obtained over a wide range in a short scanning time. Our purpose was to determine whether CT angiography using the MDCT system could effectively depict extracranial-intracranial (EC-IC) bypass routes. METHOD: Helical CT angiography was performed using an MDCT scanner in 12 patients who had undergone EC-IC bypass surgery: 10 patients had undergone superficial temporal artery-middle cerebral artery (MCA) anastomosis, 1 patient had undergone an encephaloduroarteriosynangiosis procedure for the treatment of moyamoya disease, and 1 patient had undergone an external carotid artery-MCA anastomosis using a graft. The resulting CT angiograms were visually evaluated for their depiction of the EC-IC bypass route. Conventional angiograms were available for comparison in all 12 patients. RESULTS: The EC-IC bypass was visualized to be patent at the site of anastomosis in all 12 patients. Branches of the MCA secondary or more to those connected to a donor artery were demonstrated in 9 patients, whereas MCA branches immediately distal to the anastomosis were demonstrated in 3 patients. The CT angiography findings corresponded well with the conventional angiography findings in all patients. CONCLUSION: Helical CT angiography using the MDCT technique is an effective method for visualizing EC-IC bypass routes.

MR-angiogram-added surface anatomy scanning of superficial cerebral arteriovenous malformations.

Our purpose was to evaluate the utility of surface anatomy scanning (SAS) of the brain with superimposition of MR angiograms in the diagnosis and presurgical planning of superficial cerebral arteriovenous malformations (AVMs). We performed SAS in 15 patients with superficial cerebral AVMs. Two-dimensional phase-contrast MR angiograms were then obtained in the same section and superimposed on the SAS images. The images were assessed regarding the visualization of each AVM component using a four-point grading scale (3=excellent to 0=poor). In 13 of the patients, the assessment was made comparing with conventional angiograms. The images were also compared with the surgical findings in 5 patients. The added images agreed well with the angiographic findings and demonstrated the relationships between the AVM components and the adjacent brain surface. The average scores of the feeder(s), nidus, and drainer(s) were 2.27, 2.33, and 2.67, respectively. The surgical findings also correlated well with the added images. Our technique can noninvasively demonstrate superficial AVMs along with the brain surface and provide information useful for planning surgery.