Deep Learning-based Hierarchical Brain Segmentation with Preliminary Analysis of the Repeatability and Reproducibility.

PURPOSE: We developed new deep learning-based hierarchical brain segmentation (DLHBS) method that can segment T1-weighted MR images (T1WI) into 107 brain subregions and calculate the volume of each subregion. This study aimed to evaluate the repeatability and reproducibility of volume estimation using DLHBS and compare them with those of representative brain segmentation tools such as statistical parametric mapping (SPM) and FreeSurfer (FS). METHODS: Hierarchical segmentation using multiple deep learning models was employed to segment brain subregions within a clinically feasible processing time. The T1WI and brain mask pairs in 486 subjects were used as training data for training of the deep learning segmentation models. Training data were generated using a multi-atlas registration-based method. The high quality of training data was confirmed through visual evaluation and manual correction by neuroradiologists. The brain 3D-T1WI scan-rescan data of the 11 healthy subjects were obtained using three MRI scanners for evaluating the repeatability and reproducibility. The volumes of the eight ROIs-including gray matter, white matter, cerebrospinal fluid, hippocampus, orbital gyrus, cerebellum posterior lobe, putamen, and thalamus-obtained using DLHBS, SPM 12 with default settings, and FS with the "recon-all" pipeline. These volumes were then used for evaluation of repeatability and reproducibility. RESULTS: In the volume measurements, the bilateral thalamus showed higher repeatability with DLHBS compared with SPM. Furthermore, DLHBS demonstrated higher repeatability than FS in across all eight ROIs. Additionally, higher reproducibility was observed with DLHBS in both hemispheres of six ROIs when compared with SPM and in five ROIs compared with FS. The lower repeatability and reproducibility in DLHBS were not observed in any comparisons. CONCLUSION: Our results showed that the best performance in both repeatability and reproducibility was found in DLHBS compared with SPM and FS.

Applicability of artificial intelligence-based computer-aided detection (AI-CAD) for pulmonary tuberculosis to community-based active case finding.

BACKGROUND: Artificial intelligence-based computer-aided detection (AI-CAD) for tuberculosis (TB) has become commercially available and several studies have been conducted to evaluate the performance of AI-CAD for pulmonary tuberculosis (TB) in clinical settings. However, little is known about its applicability to community-based active case-finding (ACF) for TB. METHODS: We analysed an anonymized data set obtained from a community-based ACF in Cambodia, targeting persons aged 55 years or over, persons with any TB symptoms, such as chronic cough, and persons at risk of TB, including household contacts. All of the participants in the ACF were screened by chest radiography (CXR) by Cambodian doctors, followed by Xpert test when they were eligible for sputum examination. Interpretation by an experienced chest physician and abnormality scoring by a newly developed AI-CAD were retrospectively conducted for the CXR images. With a reference of Xpert-positive TB or human interpretations, receiver operating characteristic (ROC) curves were drawn to evaluate the AI-CAD performance by area under the ROC curve (AUROC). In addition, its applicability to community-based ACFs in Cambodia was examined. RESULTS: TB scores of the AI-CAD were significantly associated with the CXR classifications as indicated by the severity of TB disease, and its AUROC as the bacteriological reference was 0.86 (95% confidence interval 0.83-0.89). Using a threshold for triage purposes, the human reading and bacteriological examination needed fell to 21% and 15%, respectively, detecting 95% of Xpert-positive TB in ACF. For screening purposes, we could detect 98% of Xpert-positive TB cases. CONCLUSIONS: AI-CAD is applicable to community-based ACF in high TB burden settings, where experienced human readers for CXR images are scarce. The use of AI-CAD in developing countries has the potential to expand CXR screening in community-based ACFs, with a substantial decrease in the workload on human readers and laboratory labour. Further studies are needed to generalize the results to other countries by increasing the sample size and comparing the AI-CAD performance with that of more human readers.

Measurement of solid size in early-stage lung adenocarcinoma by virtual 3D thin-section CT applied artificial intelligence.

An artificial intelligence (AI) system that reconstructs virtual 3D thin-section CT (TSCT) images from conventional CT images by applying deep learning was developed. The aim of this study was to investigate whether virtual and real TSCT could measure the solid size of early-stage lung adenocarcinoma. The pair of original thin-CT and simulated thick-CT from the training data with TSCT images (thickness, 0.5-1.0 mm) of 2700 pulmonary nodules were used to train the thin-CT generator in the generative adversarial network (GAN) framework and develop a virtual TSCT AI system. For validation, CT images of 93 stage 0-I lung adenocarcinomas were collected, and virtual TSCTs were reconstructed from conventional 5-mm thick-CT images using the AI system. Two radiologists measured and compared the solid size of tumors on conventional CT and virtual and real TSCT. The agreement between the two observers showed an almost perfect agreement on the virtual TSCT for solid size measurements (intraclass correlation coefficient = 0.967, P < 0.001, respectively). The virtual TSCT had a significantly stronger correlation than that of conventional CT (P = 0.003 and P = 0.001, respectively). The degree of agreement between the clinical T stage determined by virtual TSCT and the clinical T stage determined by real TSCT was excellent in both observers (k = 0.882 and k = 0.881, respectively). The AI system developed in this study was able to measure the solid size of early-stage lung adenocarcinoma on virtual TSCT as well as on real TSCT.

Two-step artificial intelligence algorithm for liver segmentation automates anatomic virtual hepatectomy.

BACKGROUND: Anatomic virtual hepatectomy with precise liver segmentation for hemilivers, sectors, or Couinaud's segments using conventional three-dimensional simulation is not automated and artificial intelligence (AI)-based algorithms have not yet been applied. METHODS: Computed tomography data of 174 living-donor candidates for liver transplantation (training data) were used for developing a new two-step AI algorithm to automate liver segmentation that was validated in another 51 donors (validation data). The Pure-AI (no human intervention) and ground truth (GT, full human intervention) data groups were compared. RESULTS: In the Pure-AI group, the median Dice coefficients of the right and left hemilivers were highly similar, 0.95 and 0.92, respectively; sectors, posterior to lateral: 0.86-0.92, and Couinaud's segments 1-8: 0.71-0.89. Labeling of the first-order branch as hemiliver, right or left portal vein perfectly matched; 92.8% of the second-order (sectors); 91.6% of third-order (segments) matched between the Pure-AI and GT data. CONCLUSIONS: The two-step AI algorithm for liver segmentation automates anatomic virtual hepatectomy. The AI-based algorithm correctly divided all hemilivers, and more than 90% of the sectors and segments.

Three-dimensional conditional generative adversarial network-based virtual thin-slice technique for the morphological evaluation of the spine.

Virtual thin-slice (VTS) technique is a generative adversarial network-based algorithm that can generate virtual 1-mm-thick CT images from images of 3-10-mm thickness. We evaluated the performance of VTS technique for assessment of the spine. VTS was applied to 4-mm-thick CT images of 73 patients, and the visibility of intervertebral spaces was evaluated on the 4-mm-thick and VTS images. The heights of vertebrae measured on sagittal images reconstructed from the 4-mm-thick images and VTS images were compared with those measured on images reconstructed from 1-mm-thick images. Diagnostic performance for the detection of compression fractures was also compared. The intervertebral spaces were significantly more visible on the VTS images than on the 4-mm-thick images (P < 0.001). The absolute value of the measured difference in mean vertebral height between the VTS and 1-mm-thick images was smaller than that between the 4-mm-thick and 1-mm-thick images (P < 0.01-0.54). The diagnostic performance of the VTS images for detecting compression fracture was significantly lower than that of the 4-mm-thick images for one reader (P = 0.02). VTS technique enabled the identification of each vertebral body, and enabled accurate measurement of vertebral height. However, this technique is not suitable for diagnosing compression fractures.

Artificial intelligence analysis of three-dimensional imaging data derives factors associated with postoperative recurrence in patients with radiologically solid-predominant small-sized lung cancers.

OBJECTIVES: Indications of limited resection, such as segmentectomy, have recently been reported for patients with solid-predominant lung cancers ≤2 cm. This study aims to identify unfavourable prognostic factors using three-dimensional imaging analysis with artificial intelligence (AI) technology. METHODS: A total of 157 patients who had clinical N0 non-small cell lung cancer with a radiological size ≤2 cm, and a consolidation tumour ratio > 0.5, who underwent anatomical lung resection between 2011 and 2017 were enrolled. To evaluate the three-dimensional structure, the ground-glass nodule/Solid Automatic Identification AI software Beta Version (AI software; Fujifilm Corporation, Japan) was used. RESULTS: Maximum standardized uptake value (SUVmax) and solid-part volume measured by AI software (AI-SV) showed significant differences between the 139 patients with adenocarcinoma and the 18 patients with non-adenocarcinoma. Among the adenocarcinoma patients, 42 patients (30.2%) were found to be pathological upstaging. Multivariable analysis demonstrated that high SUVmax, high carcinoembryonic antigen level and high AI-SV were significant prognostic factors for recurrence-free survival (RFS; P < 0.05). The 5-year RFS was compared between patients with tumours showing high SUVmax and those showing low SUVmax (67.7% vs 95.4%, respectively, P < 0.001). The 5-year RFS was 91.0% in patients with small AI-SV and 68.1% in those with high AI-SV (P = 0.001). CONCLUSIONS: High AI-SV, high SUVmax and abnormal carcinoembryonic antigen level were unfavourable prognostic factors of patients with solid-predominant lung adenocarcinoma with a radiological size ≤2 cm. Our results suggest that lobectomy should be preferred to segmentectomy for patients with these prognostic factors.

Development and external validation of a deep learning-based computed tomography classification system for COVID-19.

BACKGROUND: We aimed to develop and externally validate a novel machine learning model that can classify CT image findings as positive or negative for SARS-CoV-2 reverse transcription polymerase chain reaction (RT-PCR). METHODS: We used 2,928 images from a wide variety of case-control type data sources for the development and internal validation of the machine learning model. A total of 633 COVID-19 cases and 2,295 non-COVID-19 cases were included in the study. We randomly divided cases into training and tuning sets at a ratio of 8:2. For external validation, we used 893 images from 740 consecutive patients at 11 acute care hospitals suspected of having COVID-19 at the time of diagnosis. The dataset included 343 COVID-19 patients. The reference standard was RT-PCR. RESULTS: In external validation, the sensitivity and specificity of the model were 0.869 and 0.432, at the low-level cutoff, 0.724 and 0.721, at the high-level cutoff. Area under the receiver operating characteristic was 0.76. CONCLUSIONS: Our machine learning model exhibited a high sensitivity in external validation datasets and may assist physicians to rule out COVID-19 diagnosis in a timely manner at emergency departments. Further studies are warranted to improve model specificity.

Coronary lumen and plaque segmentation from CTA using higher-order shape prior.

We propose a novel segmentation method based on multi-label graph cuts utilizing higher-order potentials to impose shape priors. Each higher-order potential is defined with respect to a candidate shape, and takes a low value if and only if most of the voxels inside the shape are foreground and most of those outside are background. We apply this technique to coronary lumen and plaque segmentation in CT angiography, exploiting the prior knowledge that the vessel walls tend to be tubular, whereas calcified plaques are more likely globular. We use the Hessian analysis to detect the candidate shapes and introduce corresponding higher-order terms into the energy. Since each higher-order term has any effect only when its highly specific condition is met, we can add many of them at possible locations and sizes without severe side effects. We show the effectiveness of the method by testing it on the standardized evaluation framework presented at MICCAI segmentation challenge 2012. The method achieved values comparable to the best in each of the sensitivity and positive predictive value, placing it at the top in average rank.

Investigation of alternative compounds to poly(E-MA) as a polymeric surfactant for preparation of microcapsules by phase separation method.

Various water-soluble polymers were used to examine an alternative emulsifier for poly(ethylene-alt-maleic anhydride), used in the preparation of crosslinked polyurea microcapsules. Microcapsules were successfully prepared by using the water-soluble polymers with large molecular weight alternating copolymers, namely poly(olefin-maleic anhydride), poly(olefin-maleic acid), and poly(acrylic acid). On the other hand, no microcapsule resulted from olefin-maleic acid with small molecular weight alternating copolymers. From these results, the following guidelines were obtained for the selection of polymeric surfactants suitable for crosslinked polyurea microcapsule. A polymeric surfactant must have maleic acid or a carboxyl group in order to form a crosslinked polyurea microcapsule membrane. Furthermore, to form a stronger capsule membrane it is desirable to have a maleic anhydride group. It is also important for membrane formation that the polymeric surfactant has a suitable molecular weight.

Preparation and interfacial properties of a novel biodegradable polymer surfactant: poly(ethylene oxide monooleate-block-DL-lactide).

In this paper, we report a novel synthesis of poly(ethylene oxide monooleate-block-DL-lactide) (MOPEO-PLA) in the presence of stannous 2-ethylhexanoate catalyst. By utilizing the surfactant property and the reactive double bond of the amphiphilic MOPEO-PLA, various characteristics of PLA microspheres, such as surface and internal structure, surface morphology, release property, and so on, may potentially be controlled. MOPEO-PLA was found to be hydrophobic enough to prevent loss by dissolution into aqueous solution, which is often a problem for MOPEO. Furthermore, the interfacial tension measurements of a MOPEO-PLA/toluene/water system revealed that MOPEO-PLA had a good surface activity almost equal to that of MOPEO. The MOPEO-PLA/PLA blend films were prepared by solvent casting on a water layer. Contact-angle measurements of MOPEO-PLA/PLA blend films confirmed that the hydrophilic PEO segments were selectivity accumulated at the oil/water interface. Moreover, the surface free energy on the 'water side' of the MOPEO-PLA/PLA blend films was increased because of the increase in polar components as a result of the ether bonds of the PEO segments. Schematic illustration of the adsorption property of a) MOPEO-PLA with a high-molecular-weight PLA segment and b) MOPEO-PLA with a low-molecular-weight PLA segment at an ethyl acetate/water interface.

Surface morphology control of polylactide microspheres enclosing irinotecan hydrochloride.

In order to reduce the initial burst from polylactide (PLA) microspheres enclosing an antitumor agent, we prepared the microspheres with a smooth surface by varying solvent evaporation conditions such as operating temperature and pressure. PLA microspheres enclosing irinotecan hydrochloride (CPT) were prepared using the O/O emulsion system for solvent evaporation. The mean diameter and enclosing efficiency were almost constant because they were independent of solvent evaporation conditions. Scanning electron microscopic (SEM) observation verified the smooth surface of the PLA microspheres produced by varying the preparation conditions. In vitro release experiments show that the initial burst of microspheres with a smooth surface was less than that of those with a rough surface.

Bimodal fingerprint capturing system based on compound-eye imaging module.

A compact imaging system called TOMBO (Thin Observation Module by Bound Optics) is proposed in which a microlens array is used for thin hardware configuration. This paper describes a fingerprint-capturing module as an application of the TOMBO. Experimental results by the TOMBO prototype system are shown to clarify the applicability of the TOMBO to the fingerprint capturing. Different types of biometrics, i.e., fingerprint and face images, are captured by the same hardware, which shows the extendability of the system for multimodal identification.

Reconstruction of a high-resolution image on a compound-eye image-capturing system.

The authors have proposed an architecture for a compact image-capturing system called TOMBO (thin observation module by bound optics), which uses compound-eye imaging for a compact hardware configuration [Appl. Opt. 40, 1806 (2001)]. The captured compound image is decomposed into a set of unit images, then the pixels in the unit images are processed with digital processing to retrieve the target image. A new method for high-resolution image reconstruction, called a pixel rearrange method, is proposed. The relation between the target object and the captured signals is estimated and utilized to rearrange the original pixel information. Experimental results show the effectiveness of the proposed method. In the experimental TOMBO system, the resolution obtained is four times higher than that of the unit image that did not undergo reconstruction processing.

Multispectral imaging using compact compound optics.

A very thin image capturing system called TOMBO (thin observation module by bound optics) is developed with compound-eye imaging and digital post-processing. As an application of TOMBO, a multispectral imaging system is proposed. With a specific arrangement of the optical system, spatial points can be observed by multiple photodetectors simultaneously. A filter array inserted in front of the image sensor enables observation of the spectrum of the target. The captured image is reconstructed by a modified pixel rearranging method extended to treat multi-channel spectral data, in which pixels in the captured image are geometrically rearranged onto a multi-channel virtual image plane. Experimental results of the image reconstruction show the effectiveness of the proposed system.

Color imaging with an integrated compound imaging system.

Color-imaging methods with an integrated compound imaging system called TOMBO (Thin observation module by bound optics) are presented. The TOMBO is a compact optoelectronic imaging system for image capturing based on compound-eye imaging and post digital processing. First, a general description of the TOMBO system is given, and then two configurations for color imaging are described. Experimental comparison of these configurations is made by use of an experimental TOMBO system. The characteristics and the performance on the proposed methods are briefly discussed.

Mechanical properties and osteoconductivity of new bioactive composites consisting of partially crystallized glass beads and poly(methyl methacrylate).

New bioactive composites consisting of partially crystallized glass beads as inorganic fillers and poly(methyl methacrylate) (PMMA) as an organic matrix were developed. Two kinds of partially crystallized glass beads, designated Cry820 and Cry850, were newly prepared by the heating of MgO-CaO-SiO(2)-P(2)O(5) glass at 820 and 850 degrees C, respectively. The glass beads were mixed with PMMA to form two new composites designated Cry820C and Cry850C, respectively. The goal of this study was to produce a highly osteoconductive and mechanically strong composite cement with these new fillers. A previously reported composite cement designated AWC, which was composed of apatite- and wollastonite-containing glass ceramic (AW-GC) as a powder filler and the same PMMA polymer used in the new composites, was used as a reference material. The quantity of filler added to each composite was 70 wt %. The bending strength of Cry820C was significantly higher than that of Cry850C. Composites were packed into intramedullary canals of rat tibiae to evaluate their osteoconductivity, as determined by an affinity index. The affinity index, which equaled the length of bone in direct contact with the composite surface expressed as a percentage of the total length of the composite surface, was calculated for each composite. The rats were euthanized at 4, 8, and 25 weeks after implantation. At each time interval studied, Cry820C showed a significantly higher affinity index than AWC up to 25 weeks after implantation. Cry850C showed a significantly higher affinity index than AWC up to 8 weeks and a higher affinity index than AWC at 25 weeks, although the difference was not significant. The values for each composite increased significantly with time up to 25 weeks. Our study revealed that the higher osteoconductivity of the new composites was due to the larger quantity of the glassy phase of the crystallized glass beads at the composite surface and the lower solubility of the PMMA powder to methyl methacrylate monomer. In addition, the spherical shape of the crystallized glass beads gave the new composites strong enough mechanical properties to be useful under weight-bearing conditions. The new composites show promise as alternatives, with improved properties, to conventional PMMA bone cement.

Composites consisting of poly(methyl methacrylate) and alumina powder: an evaluation of their mechanical and biological properties.

We developed new composites consisting of comparatively high molecular weight poly(methyl methacrylate) (hPMMA) and delta-alumina powder or alpha-alumina powder (designated delta-APC and alpha-APC, respectively) that allowed direct bone formation on their surfaces in vivo. delta-Alumina powder was manufactured by the fusing and quenching of pulverized alumina powder. It was composed mainly of delta-crystal phases of alumina. The purpose of this study was to evaluate the static mechanical properties and biological properties of these composites. The hPMMA itself was used as a reference material. The bending strength and Young's modulus of both delta-APC and alpha-APC were significantly higher than those of hPMMA, and the alumina composites are believed to be strong enough for use under weight-bearing conditions. The three types of composites were packed into the intramedullary canals of rat tibiae to evaluate osteoconductivity, as determined by an affinity index. Rats were sacrificed 4 and 8 weeks after surgery. The affinity index, equal to the length of bone in direct contact with the composite surface and expressed as a percentage of the total length of the composite surface, was calculated for each composite at each interval. Histologically, new bone had formed along the surfaces of both delta-APC and alpha-APC within 4 weeks. The affinity indices for both delta-APC and alpha-APC increased significantly with time up to 8 weeks. At 8 weeks, the affinity index for delta-APC was significantly higher than the indices for alpha-APC and hPMMA. This study revealed that the excellent osteoconductivity of delta-APC was due to the delta-crystal phases of alumina and the high molecular weight of hPMMA. delta-APC shows promise as a base for developing a highly osteoconductive and mechanically strong biomaterial.

PMMA-based bioactive cement: effect of glass bead filler content and histological change with time.

A new bioactive bone cement (designated GBC), which is a polymethyl methacrylate (PMMA)-based composite consisting of bioactive glass beads as an inorganic filler and high molecular-weight PMMA as an organic matrix, has been developed. The purpose of the present study was to evaluate the effect of the filler content on the mechanical properties and osteoconductivity of GBC, to decide the most suitable filler proportion, and to evaluate the degree of cement degradation with time. The bioactive beads, consisting of MgO-CaO-SiO(2)-P(2)O(5)-CaF(2) glass, were added to the cement in various proportions (40-70 wt %). The bending strength of GBC did not differ among the proportions (approximately 136 MPa), but the elastic modulus of bending of GBC increased as the glass bead filler content increased (approximately 4.1-7.2 GPa). The all types of GBC were packed into the intramedullary canals of rat tibiae to evaluate osteoconductivity, as determined by an affinity index calculated as the length of bone in direct contact with the cement surface expressed as a percentage of the total length of the cement surface. Rats were sacrificed at 4, 8, 25, and 39 weeks after implantation, and the affinity index was calculated for each type of GBC at each time point. Histologically, new bone had formed along the surface of all types of GBC within 4 weeks, even in GBC containing only 40 wt % of glass beads. The affinity indices of GBC tended to increase as the proportion of glass bead filler increased and as the implantation period increased. In GBC containing 60 or 70 wt % of glass beads, significant rapid increases in the affinity indices were found from 4 to 8 weeks, and the high values (approximately 70%) were maintained up to 39 weeks. A sign of glass bead degradation was observed at the bone-cement interface in the rat tibiae at 39 weeks. We conclude that, when mechanical properties and osteoconductivity are both taken into consideration, GBC containing 60 or 70 wt % of glass beads is the most suitable formulation, but that further studies are needed to investigate and overcome the degradation.