Delineation of the intratemporal facial nerve in a cadaveric specimen on diffusion tensor imaging using a 9.4 T magnetic resonance imaging scanner: a technical note Abstract The purpose of this study was to determine whether the intratemporal facial nerve could be delineated on 9.4 T magnetic resonance imaging (MRI) using T2-weighted and diffusion tensor imaging (DTI). DTI using a b value of 3000 and an isotropic resolution of 0.4 mm3 on a 9.4 T MRI scanner was performed on a whole-block celloidin-embedded cadaveric temporal bone specimen of a 1-year-old infant with normal temporal bones. The labyrinthine, tympanic, and mastoid segments of the facial nerve and the chorda tympani nerve were readily depicted on DTI. Therefore, DTI performed using a high b value on a high-field strength MRI scanner could help evaluate the intratemporal facial nerve in whole temporal bone ex vivo specimens.

Biological polymeric shielding design for an X-ray laboratory using Monte Carlo codes Abstract Photon irradiation facilities are often shielded using lead despite its toxicity and high cost. In this study, three Monte Carlo codes, EGS5, MCNPX, and Geant4, were utilized to investigate the efficiency of a relatively new polymeric base compound (CnH2n), as a radiation shielding material for photons with energies below 150 keV. The proposed compound with the densities of 6 and 8 g cm−3 were doped with the weight percentages of 8.0 and 15.0% gadolinium. The probabilities of photoelectric effect and Compton scattering were relatively equal at low photon energies, thus the shielding design was optimized using three Monte Carlo codes for the conformity of calculation results. Consequently, 8% Gd-doped polymer with thickness less than 2 cm and density of 6 g cm−3was adequate for X-ray room shielding to attenuate more than 95% of the 150-keV incident photons. An average dose rate reduction of 88% can be achieved to ensure safety of the radiation area.

Evaluation of a 3D-printed heterogeneous anthropomorphic head and neck phantom for patient-specific quality assurance in intensity-modulated radiation therapy Abstract We evaluated an anthropomorphic head and neck phantom with tissue heterogeneity, produced using a personal 3D printer, with quality assurance (QA), specific to patients undergoing intensity-modulated radiation therapy (IMRT). Using semi-automatic segmentation, 3D models of bone, soft tissue, and an air-filled cavity were created based on computed tomography (CT) images from patients with head and neck cancer treated with IMRT. For the 3D printer settings, polylactide was used for soft tissue with 100% infill. Bone was reproduced by pouring plaster into the cavity created by the 3D printer. The average CT values for soft tissue and bone were 13.0 ± 144.3 HU and 439.5 ± 137.0 HU, respectively, for the phantom and 12.1 ± 124.5 HU and 771.5 ± 405.3 HU, respectively, for the patient. The gamma passing rate (3%/3 mm) was 96.1% for a nine-field IMRT plan. Thus, this phantom may be used instead of a standard shape phantom for patient-specific QA in IMRT.

Evaluation of a new commercial automated planning software for tangential breast intensity-modulated radiation therapy Abstract Automated treatment planning may decrease the effort required in planning and promote increased routine clinical use of intensity-modulated radiation therapy (IMRT) for many breast cancer patients. The aim of this study was to evaluate a new commercial automated planning software for tangential breast IMRT by comparing it with clinical plans from whole-breast irradiation. We prospectively enrolled 150 patients with Stage 0–1 breast cancer who underwent breast-conserving surgery at our institution between September 2016 and August 2017. Total doses of 42.56 Gy in 16 fractions (n = 98) or 50 Gy in 25 fractions (n = 44) were used. All treatment plans were retrospectively re-planned using the automated breast planning (ABP) software. All automated plans generated clinically deliverable beam parameters with no patient body collision and no contralateral breast pass through. The mean homogeneity index of the automatically generated clinical target volume, percentage volume of lungs receiving dose more than 20 Gy, mean heart dose, and dose to the highest irradiated 2-cc volumes of the irradiated volume were 0.077 ± 0.019, 4.2% ± 1.2%, 142 ± 69 cGy, and 105.8% ± 1.7% (prescribed dose: 100%), respectively. The mean planning time was 4.8 ± 1.4 min. The ABP software demonstrated high clinical acceptability and treatment planning cost efficiency for tangential breast IMRT. The ABP software may be useful for delivering high-quality treatment to a majority of patients with early-stage breast cancer.

The feasibility of contrast-enhanced spectral mammography immediately after contrast-enhanced CT Abstract Contrast-enhanced spectral mammography (CESM) is a digital mammography method that requires an intravenous injection of iodinated contrast material to detect hypervascular lesions. In patients undergoing evaluation for metastases before breast tumor surgery, a contrast material must be injected for computed tomography (CT) and CESM studies. The purpose of our study was to investigate the feasibility of performing CESM immediately after contrast-enhanced CT, without injecting additional contrast material. We enrolled 77 women with 88 breast carcinomas. Immediately after contrast-enhanced CT, we performed CESM without injecting additional contrast material. The patients were divided into two groups based on the length of the interval between contrast material injection and the start of mammography. In group A (n = 51), it was less, and in group B (n = 26) it was more than 7 min. We measured the tumor gland contrast of each tumor on the CESM images and recorded the tumor opacification on a 4-point visual scale. The mean interval between the start of contrast material injection for CT and the acquisition of mammograms in groups A and B was 5.41 and 10.40 min, respectively. All lesions were detectable on the CESM images. There was no significant difference in the visual evaluation between the two groups (p = 0.21). CESM immediately after contrast-enhanced CT without the injection of additional contrast material is feasible and cost-effective.

Overview of image-to-image translation by use of deep neural networks: denoising, super-resolution, modality conversion, and reconstruction in medical imaging Abstract Since the advent of deep convolutional neural networks (DNNs), computer vision has seen an extremely rapid progress that has led to huge advances in medical imaging. Every year, many new methods are reported at conferences such as the International Conference on Medical Image Computing and Computer-Assisted Intervention and Machine Learning for Medical Image Reconstruction, or published online at the preprint server arXiv. There is a plethora of surveys on applications of neural networks in medical imaging (see [1] for a relatively recent comprehensive survey). This article does not aim to cover all aspects of the field, but focuses on a particular topic, image-to-image translation. Although the topic may not sound familiar, it turns out that many seemingly irrelevant applications can be understood as instances of image-to-image translation. Such applications include (1) noise reduction, (2) super-resolution, (3) image synthesis, and (4) reconstruction. The same underlying principles and algorithms work for various tasks. Our aim is to introduce some of the key ideas on this topic from a uniform viewpoint. We introduce core ideas and jargon that are specific to image processing by use of DNNs. Having an intuitive grasp of the core ideas of applications of neural networks in medical imaging and a knowledge of technical terms would be of great help to the reader for understanding the existing and future applications. Most of the recent applications which build on image-to-image translation are based on one of two fundamental architectures, called pix2pix and CycleGAN, depending on whether the available training data are paired or unpaired (see Sect. 1.3). We provide codes ([2, 3]) which implement these two architectures with various enhancements. Our codes are available online with use of the very permissive MIT license. We provide a hands-on tutorial for training a model for denoising based on our codes (see Sect. 6). We hope that this article, together with the codes, will provide both an overview and the details of the key algorithms and that it will serve as a basis for the development of new applications.

Influence of image noise and object size on segmentation accuracy of FDG-PET imaging: a phantom experiment Abstract We aimed to evaluate the influence of noise and object size on segmentation accuracy of fluorodeoxyglucose positron emission tomography (FDG-PET) imaging. The scanned data of spherical phantoms were used. For the gradient method, 40% maximum standardized uptake value (SUVmax) method, and SUV of 2.5 threshold method, we evaluated the correlation between segmentation accuracy and background variability and that between segmentation accuracy and sphere diameters. For the gradient method, background variability did not affect segmentation accuracy, but sphere diameters had a small effect. As for the 40% SUVmax threshold method, both sphere diameters and background variability affected the segmentation accuracy. In the SUV of 2.5 threshold method, segmentation accuracy was affected by sphere diameters but not by background variability. With regard to segmentation accuracy of FDG-PET imaging, the gradient method may be more accurate and reliable compared to threshold methods when applied to images with varying noise or object size.

A complementary scheme for automated detection of high-uptake regions on dedicated breast PET and whole-body PET/CT Abstract In this study, we aimed to develop a hybrid method for automated detection of high-uptake regions in the breast and axilla using dedicated breast positron-emission tomography (db PET) and whole-body PET/computed tomography (CT) images. In our proposed method, high-uptake regions in the breast and axilla were detected using db PET images and whole-body PET/CT images. In db PET images, high-uptake regions in the breast were detected using adaptive thresholding technique based on the noise characteristics. In whole-body PET/CT images, the region of the breast that includes the axilla was first extracted using CT images. Next, high-uptake regions in the extracted breast region were detected on the PET images. By integration of the results of the two types of PET images, a final candidate region was obtained. In the experiments, the accuracy of extracting the region of the breast and detection ability was evaluated using clinical data. As a result, all breast regions were extracted correctly. The sensitivity of detection was 0.765, and the number of false positive cases were 1.8, which was 30% better than those on whole-body PET/CT alone. These results suggested that the proposed method, combining the two types of PET images is effective for improving detection performance.

Development of voxel-based optimization diffusion kurtosis imaging (DKI) and comparison with conventional DKI Abstract The aims of this study were to implement voxel-based optimization diffusion kurtosis imaging (DKI) and evaluate the accuracy of the method for the analysis of diffusion imaging data in comparison with conventional DKI. Conventional DKI and voxel-based optimization DKI were tested on a phantom and a human in a 1.5 T whole-body scanner. The differences in the diffusion coefficient (D) and diffusion kurtosis (K) values were analyzed using the Mann–Whitney U test, and the Holm correction was applied to the statistical analyses. In the phantom study, the D value resulting from voxel-based optimization DKI was significantly lower than those from conventional DKI in water and agarose solutions at concentrations of 50 and 100 g/L (all p < 0.01). Moreover, the K value was significantly lower in the water and agarose solutions at concentrations of 50, 100, and 200 g/L (all p < 0.01). In the human study, the D value resulting from voxel-based optimization DKI was significantly lower than that of conventional DKI in both white matter (WM), and gray matter (GM) (all p < 0.01). Moreover, the K value was significantly lower in cerebrospinal fluid, WM, and GM (all p < 0.01). To correctly measure the DKI, the optimized b values for each voxel must be used. Voxel-based optimization DKI is a method that optimizes the b values for each voxel. It appears that voxel-based optimization DKI improves the accuracy of the K value for biological tissues.

Determination of appropriate conversion factors for calculating size-specific dose estimates based on X-ray CT scout images after miscentering correction Abstract In this study, we proposed and evaluated the validity of an optimized size-specific dose estimate, a widely used index of radiation dose in X-ray computed tomography (CT) examinations. Based on miscentering correction of scout images, we determined the appropriate conversion factors (CF) by using a phantom. Scans were conducted using a multi-detector CT system (Aquilion ONE, Canon Medical Systems). Four cylindrical phantoms were taken in the anteroposterior (AP) and axial directions to determine the relationship between pixel value and water-equivalent length (Lw). In the AP scout image, the pixel values at the selected slice positions were converted to Lw to calculate the water-equivalent diameter (Dw). The CF was derived from Dw and CF values before and after miscentering correction was calculated. Finally, the CF values were compared to those calculated from the axial image using the conventional methodology of the American Association of Physicists in Medicine. Before miscentering correction, the maximum difference between the CF values of the axial and scout images was 7.26%. However, after miscentering correction, the maximum difference was 1.34%. Validation using a whole-body phantom generally revealed low maximum differences between the CF from the axial image and the values from the miscentering-corrected scout images. These were 2.41% in the chest, 6.30% in the upper abdomen, 1.43% in the abdomen, and 2.45% in the pelvic region. Consequently, we concluded that our miscentering correction method for deriving the appropriate CF values based on scout images is advantageous.