Effect of Contrast Level and Image Format on a Deep Learning Algorithm for the Detection of Pneumothorax with Chest Radiography.

Under the black-box nature in the deep learning model, it is uncertain how the change in contrast level and format affects the performance. We aimed to investigate the effect of contrast level and image format on the effectiveness of deep learning for diagnosing pneumothorax on chest radiographs. We collected 3316 images (1016 pneumothorax and 2300 normal images), and all images were set to the standard contrast level (100%) and stored in the Digital Imaging and Communication in Medicine and Joint Photographic Experts Group (JPEG) formats. Data were randomly separated into 80% of training and 20% of test sets, and the contrast of images in the test set was changed to 5 levels (50%, 75%, 100%, 125%, and 150%). We trained the model to detect pneumothorax using ResNet-50 with 100% level images and tested with 5-level images in the two formats. While comparing the overall performance between each contrast level in the two formats, the area under the receiver-operating characteristic curve (AUC) was significantly different (all p < 0.001) except between 125 and 150% in JPEG format (p = 0.382). When comparing the two formats at same contrast levels, AUC was significantly different (all p < 0.001) except 50% and 100% (p = 0.079 and p = 0.082, respectively). The contrast level and format of medical images could influence the performance of the deep learning model. It is required to train with various contrast levels and formats of image, and further image processing for improvement and maintenance of the performance.

Therapeutic effect of local photothermal heating of gold nanoparticle-coated self-expandable metallic stents for suppressing granulation tissue formation in the mouse colon.

PURPOSE: To investigate the therapeutic effect of local photothermal (PT) heating on suppression of stent-induced granulation tissue formation in mouse colon. MATERIALS AND METHODS: A gold nanoparticle (GNP)-coated self-expandable metallic stent (SEMS) was prepared using a two-step synthesis process for local PT heating under near-infrared laser irradiation. Twenty-four mice were randomly divided into two groups of 12 and subjected to SEMS placement in the colon. Group A received a GNP-coated SEMS without local heating and Group B received a GNP-coated SEMS and underwent local heating at 55°C after SEMS placement. The therapeutic effect of local heating was assessed by comparing the histopathological, immunohistochemical, and endoscopic results. RESULTS: Four mice were excluded because of stent migration (n = 3, group B) or death (n = 1, group A). Stent-induced granulation tissue-related variables were significantly lower in group B than in group A (p < 0.001). In vivo endoscopic images, 4 weeks after stent placement, showed granulation tissue formation over the wire mesh in group A and relatively good patency of the stented colon with no definite irregularities in group B. There was more vascular endothelial growth factor (VEGF) positivity in group A than in group B. CONCLUSION: Local PT heating suppresses granulation tissue formation after stent placement in mouse colon.

Dose-Decreasing Effect of the First Reversed Laser Beam Collimator for C-Arm Type Angiographic Equipment.

This is a study on the dose-decreasing effect of the first reversed laser beam collimator (RLBC) for C-arm type angiographic equipment. A laser beam was located at the center of each plane at an oblique angle to the angiographic equipment detector. A field of view, which could be seen with the naked eye, was made by focusing the laser beam in the direction of the X-ray source. The height of the table was fixed at 75 cm and the iron balls were located within 2 mm of the top, bottom, left, and right edges of the output image. The time needed for location fixing, fluoroscopy, and measurement of dose area product (DAP) were compared by having 30 radiologists perform location fixing by looking at the fluoroscopic image while performing location fixing (no radiation) and while the RLBC was turned on. In the next test, the time needed for location fixing, fluoroscopy, and DAP were compared when varying the location of the iron balls from 2 to 10 mm from the edges of the output image. The results showed that the time needed for location fixing, the time needed for fluoroscopy, and DAP decreased, both in the first test and the second test. This study confirmed that the use of a RLBC for C-arm type angiographic equipment decreases both the time needed to perform the procedure and the radiation dose received. It is expected that continuous advancement of RLBC technology will contribute greatly to decreasing the dose of radiation needed and improving convenience during angiography.

Patient dose for computed tomography examination: dose reference levels and effective doses based on a national survey of 2013 in Korea.

14,620 sets of patient dose data were obtained for 31 different models of computed tomography (CT) equipment (total 73) with 18 types of CT examination in Korea. Specific diagnostic reference levels (DRLs) for this study in terms of third quartile volumetric CT dose index in mGy [and dose-length product (DLP) in mGy.cm] are as follows: head, 53 (910); neck, 20 (770); chest, 14 (710); abdomen, 14 (1000); stomach, 14 (1000); liver, 14 (1700); pancreas, 14 (1700); kidney, 14 (2100); cervical spine, 30 (600); lumbar spine, 25 (760); hip, 17 (600); cardiac CT angiography, 45 (1250); head CT angiography, 43 (1900); liver CT angiography, 14 (1400) and thoraco-abdominal CT angiography, 16 (2000). In the present study, DRLs in terms of volumetric CT dose index were below previously published reference levels, partly because the newer CT equipments have improved technology that facilitates lower patient dose. Meanwhile, DRLs in terms of DLP were higher, because multi-phase scanning protocols with prolonged scan coverage have been widely used.

Effectiveness of a silicone device for foot MRI in order to obtain homogeneous fat suppression images.

BACKGROUND: Incomplete fat suppression induced by magnetic field inhomogeneity is difficult to compensate for with hardware magnetic-field shimming. PURPOSE: To evaluate the effectiveness of a silicone device used to obtain homogeneous fat suppression during 3T magnetic resonance imaging (MRI) scans of the foot. MATERIAL AND METHODS: Thirty-eight healthy volunteers were enrolled and examined twice, before (group A) and after (group B) the application of a silicone device. Fat-saturated, T2-weighted, fast spin-echo images were acquired using the same scanning protocol at both examinations. Signal- and contrast-to-noise ratios (SNR and CNR) were calculated and compared in the four regions of interest (ROIs). ROI 1 and 2 were selected from toe-side bone and soft tissue, while ROI 3 and 4 were selected from proximal bone and soft tissue. Qualitative analysis using a four-point scale was performed for three categories. The categories are as follows: the overall image quality, homogeneity of the first phalange and metatarsal bone, respectively. RESULTS: The SNR and CNR in ROI 1 and 2 were significantly higher in group A than in group B (SNR; P < 0.001, CNR; P < 0.001), and there were no significant difference in ROI 3 and 4. The qualitative score of the overall fat suppression in group B was significantly higher than that in group A (P < 0.001). Homogeneity of the first phalange in group B was also significantly higher than that in group A (P < 0.001). On the other hand, the homogeneity of the metatarsal bone was not significantly different in the two groups. CONCLUSION: The use of a silicone device provides homogeneous fat suppression in 3T MRI of the foot and can significantly improve image quality.

Radiation exposure in multidetector CT: dose comparison between late 1990s and early 2010s in Korea.

The effective dose under ordinary clinical computed tomography (CT) protocols using three kinds of 64-channel and a 40-channel CT ranged from 0.6 to 15.5 mSv in early 2010s. And the organ dose ranged from 1.6 to 130.4 mGy: orbital and brain doses for brain stroke CT were the highest. For the comparison of the effective dose between late 1990s and early 2010s, multidetector CT (MDCT) for high-resolution lung CT was 2.4 times higher than that of single-detector CT (SDCT) and the ratio was the highest. However, the effective dose at MDCT was 20 % lower than that of SDCT in chest CT due to applying dose-saving techniques. In organ dose comparisons, high-resolution lung CT at MDCT was 3.5-4.5 times higher than that of SDCT, and showed 1.1-1.5 times higher than that of SDCT in the head and chest CT. For the abdomen CT, the primary organ dose at MDCT was ∼30 % lower than that of SDCT.

Wheelchair type biomedical system with event-recorder function.

The present study is about a biometric system for a wheelchair, which can measure both bio-signal (ECG-Electrocardiogram, BCG-Ballistocardiogram) and kinetic signal (acceleration) simultaneously and send the data to a remote medical server. The equipment was developed with the object of building a system that measures the bio-signal and kinetic signal of a subject who is moving or at rest on a wheelchair and transmits the measured signals to a remote server through a CDMA (Code Division Multiple Access) network. The equipment is composed of body area network and remote medical server. The body area network was designed to obtain bio-signal and kinetic signal simultaneously and, on the occurrence of an event, to transmit data to a remote medical server through a CDMA network. The remote medical server was designed to display event data transmitted from the body area network in real time. The performance of the developed system was evaluated through two experiments. First, we measured battery life on the occurrence of events, and second, we tested whether biometric data are transmitted accurately to the remote server on the occurrence of an event. In the first experiment using the developed equipment, events were triggered 16 times and the battery worked stably for around 29 hours. In the second experiment, when an event took place, the corresponding data were transmitted accurately to the remote medical server through a CDMA network. This system is expected to be usable for the healthcare of those moving on a wheelchair and applicable to a mobile healthcare system.

Performance evaluation of biosignal measurement at the wheelchair system.

The purpose of this study is to measure both ECG and BCG(Ballistocariograph) signal of a subject on moving or resting wheelchair and detect the heart rate and respiratory rate and transmit an event message to remote server on emergent situation. To acquire ECG and BCG data, amplifier circuits were composed to be suitable for their characteristics. 3-axial accelerometer was built in the developed device to measure the mechanical noise that can be generated on moving wheelchair.The output signals were converted to digital data and stored in bio-signal archiving media(SD card). CDMA module was used to transmit the event data on ECG electrode detachment and the received data was monitored by the developed C# application program. 8 volunteers participated in the experiment to evaluate the validity of the developed device. When the event occurs in each subject, 48 Kbyte data, stored for 32 seconds from that point, was transmitted to remote server through CDMA cellular phone network correctly. The received data of ECG , BCG, and 3-axial acceleration could be archived in server and the heart rate and respiratory rate could be measured and analyzed. The correlation coefficients of respiratory rate in resting and moving with the real value were 0.9636 and 0.9237, respectively. The correlation coefficient ofR-R intervals between the developed and reference device was 0.999.In conclusion, the developed device in this study could acquire the ECG and BCG data of subjects on wheelchair simultaneously and measure their heart rate and respiratory rate. In addition, event data was verified to be transmitted to remote server without any errors.