Hoen-oh Shin

OncoTREAT: a software assistant for cancer therapy monitoring

AbstractObjectCancer is one of the leading causes of death worldwide and therapy options are often associated with severe stress for the patient and high costs. Therefore, precise evaluation of therapy success is essential. Material and Methods In the framework of the VICORA research project (Virtual Institute for Computer Assistance in Clinical Radiology), a software application was developed to support the radiologist in evaluating the response to tumor therapy. The application provides follow-up support for oncological therapy monitoring by volumetric quantification of lung, liver and brain metastases as well as enlarged lymph nodes and assists the user by temporal registration of lesion positions. Results With close cooperation between computer scientists and radiologists the application was tested and optimized to achieve a high degree of usability. Several clinical studies were carried out to evaluate the robustness and reproducibility of the volumetry methods. Conclusion Automatic volumetry and segmentation allows reliable detection of tumor growth and has the potential to increase reliability and significance of monitoring tumor growth in follow-up examinations.

Informatics in Radiology: Sliding-Thin-Slab Averaging for Improved Depiction of Low-Contrast Lesions with Radiation Dose Savings at Thin-Section CT

Current multidetector computed tomography (CT) scanners allow volumetric data acquisition with thin-section collimations and overlapping section reconstructions. The resultant nearly isotropic data sets help minimize partial-volume averaging effects and are ideal for two- and three-dimensional postprocessing and software-assisted lesion detection and quantification. However, the section thickness, image noise, and radiation dose are closely related, and when one parameter must be altered to suit the clinical setting, the others may be affected. When the clinical purpose demands both high spatial resolution and low image noise (eg, for the detection of hypoattenuating lesions in organs such as the kidneys and liver), the necessary trade-off--an increase in the radiation dose to the patient--may be unacceptable. The application of a sliding-thin-slab averaging algorithm during image postprocessing and review helps overcome this limitation by reconstructing thicker sections with lower noise levels from thin-section data obtained with dose-saving protocols. In principle, a high noise level is acceptable in the initial reconstruction of the CT volume data set. During image review at the workstation, the section thickness can be interactively increased to minimize image noise and improve lesion detectability. The combination of thin-section scanning with thick-section display allows routine volumetric imaging without a general increase in radiation dose or a reduction in the detectability of low-contrast lesions. Supplemental material available at http://radiographics.rsna.org/lookup/suppl/doi:10.1148/rg.302096007/-/DC1.

Repair of a segmental long bone defect in human by implantation of a novel multiple disc graft

Large segmental defects of the weight bearing long bones are very difficult to reconstruct. Current treatment options are afflicted with several limitations and disadvantages. We describe a novel approach to regenerate a segmental long bone defect in a patient using a multiple disc graft. Decellularized bovine trabecular bone discs were seeded with autologous bone marrow cells and cultured in a perfusion chamber for three weeks. Multiple cell-seeded discs were implanted to close a 72 mm defect of the distal tibia in a 58-year-old woman, and fixed by an intramedullary nail. Bone formation was assessed non-invasively by plain radiographs and 18F-labeled sodium fluoride-based co-registration of positron emission- and computed tomography (PET/CT). Bone was actively formed around the grafted defect as early as six weeks after surgery. Because the tibia was sufficiently stabilized, the patient was able to freely walk with full weight bearing 6 weeks after surgery. The uneventful two-year follow-up and the satisfaction of the patient demonstrated the success of the procedure. Therefore the use of multiple cell-seeded disc grafts can be considered as a treatment alternative for patients with segmental long bone defects.

Influence of sinogram affirmed iterative reconstruction of CT data on image noise characteristics and low-contrast detectability: an objective approach.

Objectives To utilize a novel objective approach combining a software phantom and an image quality metric to systematically evaluate the influence of sinogram affirmed iterative reconstruction (SAFIRE) of multidetector computed tomography (MDCT) data on image noise characteristics and low-contrast detectability (LCD). Materials and Methods A low-contrast and a high-contrast phantom were examined on a 128-slice scanner at different dose levels. The datasets were reconstructed using filtered back projection (FBP) and SAFIRE and virtual low-contrast lesions (-20HU) were inserted. LCD was evaluated using the multiscale structural similarity index (MS-SIM*). Image noise texture and spatial resolution were objectively evaluated. Results The use of SAFIRE led to an improvement of LCD for all dose levels and lesions sizes. The relative improvement of LCD was inversely related to the dose level, declining from 208%(±37%), 259%(±30%) and 309%(±35%) at 25mAs to 106%(±6%), 119%(±9%) and 123%(±8%) at 200mAs for SAFIRE filter strengths of 1, 3 and 5 (p<0.05). SAFIRE reached at least the LCD of FBP at a relative dose of 50%. There was no statistically significant difference in spatial resolution. The use of SAFIRE led to coarser image noise granularity. Conclusion A novel objective approach combining a software phantom and the MS-SSIM* image quality metric was used to analyze the detectability of virtual low-contrast lesions against the background of image noise as created using SAFIRE in comparison to filtered back-projection. We found, that image noise characteristics using SAFIRE at 50% dose were comparable to the use of FBP at 100% dose with respect to lesion detectability. The unfamiliar imaging appearance of iteratively reconstructed datasets may in part be explained by a different, coarser noise characteristic as demonstrated by a granulometric analysis.

Optimization of low-contrast detectability in thin-collimated modern multidetector CT using an interactive sliding-thin-slab averaging algorithm.

Objectives:To analyze the effects of the sliding-thin-slab averaging algorithm on low-contrast performance in MDCT imaging and to find reasonable parameters for clinical routine work. Materials and Methods:A low-contrast phantom simulating hypodense lesions (20 HU object contrast) was scanned with a 16-slice spiral CT scanner using different mAs-settings of 25, 50, 100, and 195 mAs. Other scan parameters were as follows: tube voltage = 120 kVp, slice collimation = 0.625 mm, pitch = 1.375 (high speed), reconstruction interval = 0.5 mm. Images were reconstructed with soft, standard, and bone algorithms, resulting in a total of 12 datasets. A sliding-thin-slab averaging algorithm was applied to these primary datasets, systematically varying the slab thickness between 0.5 and 5.0 mm. The low-contrast performance of the resulting datasets was semi-automatically analyzed using a statistical reader-independent approach: A size-dependent analysis of the image noise within the phantom was used to empirically generate a contrast discrimination function (CDF). The ratio between the actual contrast and the minimum contrast necessary for the detection (as given by the CDF) was calculated for all lesions in each dataset and used to evaluate the low-contrast detectability of the different lesions at increasing slab thickness. The results were compared with the original datasets to calculate the improvement in low-contrast detectability. Results:Using the sliding-thin-slab algorithm, low-contrast performance was increased by a factor between 1.1 and 1.7 when compared with the primary dataset. The improvement of the visibility index at optimal slab thickness when compared with the original slice thickness (0.625 mm) was statistically significant (P < 0.05, Student t test) for the following datasets: 8 mm: all datasets; 6 mm: 25 mAs/soft, 195 mAs/bone, 25 mAs/bone; 5 mm: 25 mAs/soft, 25 mAs/bone. The ideal slab thickness over all datasets was 43% (±3%) of the diameter of the lesion to be detected. Conclusions:The use of an interactive sliding-thin-slab averaging algorithm can be readily applied to optimize low-contrast detectability in thin-collimated CT datasets. As a general rule for daily routine, a slice thickness of approximately 2.5 to 3.0 mm can be regarded as a reasonable preset, resulting in an optimized detectability of lesions with a diameter of 5 mm and above.

Semiautomated segmentation of pleural effusions in MDCT datasets.

RATIONALE AND OBJECTIVES To develop and evaluate a novel algorithm for semiautomated segmentation and volumetry of pleural effusions in multidetector computed tomography (MDCT) datasets. MATERIALS AND METHODS A seven-step algorithm for semiautomated segmentation of pleural effusions in MDCT datasets was developed, mainly using algorithms from the ITK image processing library. Semiautomated segmentation of pleural effusions was performed in 40 MDCT datasets of the chest (males = 22, females = 18, mean age: 56.7 +/- 19.3 years). The accuracy of the semiautomated segmentation as compared with a manual segmentation approach was quantified based on the differences of the segmented volumes, the degree of over-/undersegmentation, and the Hausdorff distance. The time needed for the semiautomated and the manual segmentation process were recorded and compared. RESULTS The mean volume of the pleural effusions was 557.30 mL (+/- 477.27 mL) for the semiautomated and 553.19 (+/- 473.49 mL) for the manual segmentation. The difference was not statistically significant (Student t-test, P = .133). Regression analysis confirmed a strong relationship between the semiautomated algorithm and the gold standard (r(2) = 0.998). Mean overlap of the segmented areas was 79% (+/- 9.3%) over all datasets with moderate oversegmentation (22% +/- 9.3%) and undersegmentation (21% +/- 9.7%). The mean Hausdorff distance was 17.2 mm (+/- 8.35 mm). The mean duration of the semiautomated segmentation process with user interaction was 8.4 minutes (+/- 2.6 minutes) as compared to 32.9 minutes (+/- 17.4 minutes) for manual segmentation. CONCLUSION The semiautomated algorithm for segmentation and volumetry of pleural effusions in MDCT datasets shows a high diagnostic accuracy when compared with manual segmentation.

Development and first clinical application of automated virtual reconstruction of unilateral midface defects

PURPOSE Computer-assisted surgery is used for decision making, treatment, and quality control throughout the reconstruction process of unilateral midface defects. The current approaches exploit the symmetry of the face by mirroring the intact side on the defect side using various segmentation methods. All commercially available implementations, however, are somewhat time consuming and dependent on the level of expertise of the user. We present a method for automatic reconstruction of unilateral midface defects using registration. MATERIAL AND METHODS To reconstruct a skull by registration, the defect volume has to be virtually deleted from the skull. This modified data set is then mirrored and registered onto the original, defect-free skull. The fusion of these two skulls is the virtual reconstructed skull bridging the defect. Reconstruction by registration was performed for 24 different skulls without motion or dental restoration artifacts. Subsequently, simulation was performed with four accurately defined, various-sized, defects of the orbito-zygomatic complex. The results of the automated virtual reconstructions were compared with those obtained for the same defects as determined using conventional atlas-based planning software (iPlan). To simulate various clinical situations, four groups each containing six skulls were evaluated: the complete skull, midface and neurocranium, midface and lower jaw, and midface alone. The differences were compared using the similarity coefficients of Sørensen-Dice and Jaccard. Statistical analyses were performed using the t-test and Mann-Whitney U test. RESULTS The reconstruction results were similar for all the groups. The Sørensen-Dice coefficients of similarity for all reconstructed skulls were 0.869 and 0.874 for the registration and atlas-based reconstructions, respectively. The corresponding Jaccard coefficients were 0.774 and 0.781, respectively. Atlas-based reconstruction showed significantly better results in group 3 (midface and lower jaw) alone. CONCLUSION Virtual automated reconstruction by registration had equivalent accuracy to conventional atlas-based reconstruction across a spectrum of defects, from simple orbital to complex orbito-zygomatic defects. However, for those involving the midface and lower jaw, atlas-based reconstruction showed significantly better results. Although the new approach is somewhat hardware demanding, it is user independent, dispensing with the need for time-consuming adjustments to the results of planning. The first clinical application of registration reconstruction revealed performance equivalent to that of the conventional approach.

Precise determination of aortic length in patients with aortic stent grafts: in vivo evaluation of a thinning algorithm applied to CT angiography data

Abstract The aim of this study was to develop a technique for precise determination of the aortic length using volumetric CT data for potential use prior to endovascular stent-graft placement. The study population consisted of 20 patients (38 measurements) with already performed endoluminal grafting. This allowed for in vivo evaluation of our technique. Its length according to the graft specifications served as a gold standard for our own measurements. The implemented graft length varied between 120 and 195 mm. Computed tomography angiography was performed with 3-mm slice collimation, 5-mm table feed and a reconstruction interval of 2 mm. Following semi-automatic segmentation of the aorta and its large side branches, the median centerline (skeleton) of the vessels was determined employing a modified three-dimensional thinning algorithm. The algorithm was validated by comparing the calculated length of the resulting skeleton with the specifications of the grafts. The calculated length was sufficiently precise despite the limiting reconstruction interval of 2 mm of our CT data which only permitted an assessment of stent length in 2-mm steps. The differences in the measured length and graft length were in the range between 0 and 8 mm ( < 5 %) with a mean fractional error of 2.46 ± 2.37 mm. The use of an intelligent region growing algorithm capable of coping with variable arterial enhancement significantly reduced operator post-processing time. The average time necessary for segmentation was 7 min (range 3–10 min). Our algorithm provides a non-invasive method for objective and precise measurement of aortic length apparently even in tortuous vessels. It has the potential to replace angiography for aortic and iliac length measurements with calibrated catheters prior to endovascular intervention.

CT angiography for pulmonary embolism detection: the effect of breathing on pulmonary artery enhancement using a 64-row detector system.

Background Computed tomography pulmonary angiography (CTPA) is used most often in routine clinical practice for the assessment of a suspected pulmonary embolism. The diagnostic accuracy relies on sufficient contrast enhancement. Purpose: To evaluate whether image acquisition during shallow breathing can improve the image quality in patients with insufficient contrast enhancement during breath-hold examinations. Material and Methods A total of 2786 CT pulmonary angiographies, acquired on a 64-row CT during deep-inspiration breath-hold, were reviewed. Twenty-four examinations were considered non-diagnostic due to poor contrast enhancement in the pulmonary arteries (PA), although they showed preserved vascular enhancement of the superior vena cava (SVC) and the ascending aorta (AO). Eleven flawed CTPA examinations, including severe breathing artifacts and incorrect triggering were excluded. In 13 of the remaining patients, the examination was repeated during shallow breathing. Vascular contrast enhancement was compared between both scans by measuring the relative enhancement within the SVC, the main PA, and the AO. Image quality was scored by two, clinically experienced radiologists. The values are given as median and [25th;75th] quartile. Results There was a significant difference in the CT values for the PA between the repeated scans (P = 0.0002, Wilcoxon test), and with the CTPA in deep-inspiration showing a median enhancement of 97 HU (59–173), compared with 303 HU (239–385) in the CTPA acquired during free breathing. The differences for both the AO (P = 0.54) and the SVC (P = 0.78) were not significant. Scoring for the attenuation quality rose significantly (P = 0.0002) and no severe motion artifacts were detected on either scans. Conclusion If there is insufficient pulmonary artery enhancement during CTPA, attenuation of the pulmonary arteries can be improved by acquisition during shallow breathing and is without significant loss of the overall diagnostic image quality.

Imaging of the brain in polytraumatized patients comparing 64-row spiral CT with incremental (sequential) CT

INTRODUCTION Up to now, due to a better image quality, for brain imaging the substantially slower sequential examination mode has been preferred during CT in polytraumatized patients. We aimed to re-evaluate modern ultrafast 64-row spiral CT regarding image quality in brain imaging of polytraumatized patients. METHODS In 30 polytraumatized patients, both 64-row spiral and sequential CT of the brain were performed within 24h. Retrospectively, two radiologists subjectively evaluated the delineation of the internal capsule, the pons, the medial rectus muscle of the orbita, the differentiation of grey/white matter, and the extent of artifacts at the inner skull. Image noise was also evaluated objectively. Statistics were performed using Cohens kappa and a two-sided t-test. RESULTS Perfect or clear agreements were noted regarding the delineation of the inner skull, the medial rectus muscle, the internal capsule, and grey/white matter differentiation. Due to beam hardening artifacts at the level of the pons, no agreement and no superiority of one of the CT-methods was noted. No differences were obtained regarding the objective evaluation of image noise. DISCUSSION Image quality is generally equivalent. Since 64-row spiral CT can substantially save examination time we recommend to perform a spiral examination of the brain in polytraumatized patients.