Machine learning combined with radiomics and deep learning features extracted from CT images: a novel AI model to distinguish benign from malignant ovarian tumors.

BACKGROUND: To develop an artificial intelligence (AI) model with radiomics and deep learning (DL) features extracted from CT images to distinguish benign from malignant ovarian tumors. METHODS: We enrolled 149 patients with pathologically confirmed ovarian tumors. A total of 185 tumors were included and divided into training and testing sets in a 7:3 ratio. All tumors were manually segmented from preoperative contrast-enhanced CT images. CT image features were extracted using radiomics and DL. Five models with different combinations of feature sets were built. Benign and malignant tumors were classified using machine learning (ML) classifiers. The model performance was compared with five radiologists on the testing set. RESULTS:  Among the five models, the best performing model is the ensemble model with a combination of radiomics, DL, and clinical feature sets. The model achieved an accuracy of 82%, specificity of 89% and sensitivity of 68%. Compared with junior radiologists averaged results, the model had a higher accuracy (82% vs 66%) and specificity (89% vs 65%) with comparable sensitivity (68% vs 67%). With the assistance of the model, the junior radiologists achieved a higher average accuracy (81% vs 66%), specificity (80% vs 65%), and sensitivity (82% vs 67%), approaching to the performance of senior radiologists. CONCLUSIONS:  We developed a CT-based AI model that can differentiate benign and malignant ovarian tumors with high accuracy and specificity. This model significantly improved the performance of less-experienced radiologists in ovarian tumor assessment, and may potentially guide gynecologists to provide better therapeutic strategies for these patients.

Evaluating the contact anatomy and contact bone volume of spinal screws using a novel drilled surface image.

Intraoperative navigation systems have been widely applied in spinal fusion surgery to improve the implantation accuracy of spinal screws using orthogonal tomographic and surface-rendering imaging. However, these images contain limited anatomical information and no information on bone volume contact by the implanted screw, which has been proven to affect the stability of implanted screws. This study proposed a novel drilled surface imaging technique that displays anatomical integration properties to calculate the contact bone volume (CBV) of the screws implanted along an implantation trajectory. A cylinder was used to represent the area traversed by the screws, which was manually rotated and translated to a predetermined implantation trajectory according to a vertebra model obtained using computed tomography (CT) image volumes. The drilled surface image was reconstructed by interpolating the CT numbers at the predefined sampling points on the cylinder surface. The anatomical integration property and CBV of the screw implanted along the transpedicular trajectory (TT) and cortical bone trajectory (CBT) were evaluated and compared. The drilled surface image fully revealed the contact anatomical structure of the screw under the trajectories, improving the understanding of the anatomical integration of the screw and surrounding tissues. On average, the CBV of the CBT was 30% greater than that of the TT. The proposed drilled surface image may be applied in preoperative planning and integrated into intraoperative navigation systems to evaluate the anatomical integration and degree of bone contact of the screw implanted along a trajectory.

Osteogenic and angiogenic potentials of the cell-laden hydrogel/mussel-inspired calcium silicate complex hierarchical porous scaffold fabricated by 3D bioprinting.

3D printing has been popularly used in the bone tissue engineering, as many of the biomaterials for this field of study can be prepared for and produced from this additive manufacturing technique. In this study, we strategized a solvent-free processing to fabricate the polydopamine-modified calcium silicate (PDACS)/poly-caprolactone (PCL) scaffold with Wharton's jelly mesenchymal stem cells (WJMSCs) incorporated with human umbilical vein endothelial cells (HUVEC)-laden hydrogel. The PDACS/PCL/hydrogel 3D scaffold yielded a Young's modulus of the 3D scaffolds as high as 75 MPa. In addition, the vascular morphogenesis and cellular behaviors regulated by our hybrid scaffolds were also intricately evaluated. Furthermore, the HUVEC in the bioink exhibited higher levels of angiogenic biomarkers and showed potential for the formation of complex vascular networks. Higher levels of bone formation proteins were also observed in our composites. Such a hybrid of synthetic materials with cell constituents not only enhances osteogenesis but also stimulates vessel network development in angiogenesis, presenting the fact that 3D printing can be further applied in improving bone tissue regeneration in numerous aspects. We believe that this method may serve as a useful and effective approach for the regeneration of defective complex hard tissues in deep bone structures.

The Development of Skull Prosthesis Through Active Contour Model.

Skull defects result in brain infection and inadequate brain protection and pose a general danger to patient health. To avoid these situations and prevent re-injury, a prosthesis must be constructed and grafted onto the deficient region. With the development of rapid customization through additive manufacturing and 3D printing technology, skull prostheses can be fabricated accurately and efficiently prior to cranioplasty. However, an unfitted skull prosthesis made with a metal implant can cause repeated infection, potentially necessitating secondary surgery. This paper presents a method of creating suitably geometric graphics of skull defects to be applied in skull repair through active contour models. These models can be adjusted in each computed tomography slice according to the graphic features, and the curves representing the skull defect can be modeled. The generated graphics can adequately mimic the natural curvature of the complete skull. This method will enable clinical surgeons to rapidly implant customized prostheses, which is of particular importance in emergency surgery. The findings of this research can help surgeons provide patients with skull defects with treatment of the highest quality.

Polymer gel dosimeters for pretreatment radiotherapy verification using the three-dimensional gamma evaluation and pass rate maps.

Polymer gel dosimeters (PGDs) have been widely studied for use in the pretreatment verification of clinical radiation therapy. However, the readability of PGDs in three-dimensional (3D) dosimetry remain unclear. In this study, the pretreatment verifications of clinical radiation therapy were performed using an N-isopropyl-acrylamide (NIPAM) PGD, and the results were used to evaluate the performance of the NIPAM PGD on 3D dose measurement. A gel phantom was used to measure the dose distribution of a clinical case of intensity-modulated radiation therapy. Magnetic resonance imaging scans were performed for dose readouts. The measured dose volumes were compared with the planned dose volume. The relative volume histograms showed that relative volumes with a negative percent dose difference decreased as time elapsed. Furthermore, the histograms revealed few changes after 24h postirradiation. For the 3%/3mm and 2%/2mm criteria, the pass rates of the 12- and 24-h dose volumes were higher than 95%, respectively. This study thus concludes that the pass rate map can be used to evaluate the dose-temporal readability of PGDs and that the NIPAM PGD can be used for clinical pretreatment verifications.

Converting computed tomography images into photon interaction coefficients by using stoichiometric calibration and parametric fit models.

PURPOSE: X ray and γ-ray are widely applied in radiology, radiotherapy, and nuclear medicine. Linear attenuation coefficients and linear energy absorption coefficients are essential for dose calculation and image correction. In this study, a method that entails combining the stoichiometric calibration and parametric physical models was developed to convert computed tomography (CT) images into the linear attenuation coefficients and linear energy absorption coefficients. METHODS: A calibration scan was performed using standard tissue-equivalent materials to obtain the characteristics of the x-ray energy spectrum. Subsequently, relationships between CT numbers and tissue parameters were established using standard soft tissue and bone tissue data adopted from the literature. The linear attenuation coefficient and linear energy absorption coefficient were calculated using the parametric fit model. RESULTS: The results showed a linear relationship between CT numbers and tissue parameters. The tissue-equivalent materials differed from real human tissues, leading to considerable errors in estimation of mass attenuation coefficients when the photon energy was lower than 50 keV. Mass attenuation coefficients and mass energy transfer coefficients of five tissues were calculated and validated using clinical CT images. The error was less than ± 5% and ± 8%, compared with the values of the International Commission on Radiation Units (ICRU) 46 report. CONCLUSIONS: The probability of photon interaction with tissues and physical characteristics of tissues can be accurately evaluated by using the proposed method and applied in various clinical applications.

A Comprehensive Evaluation of NIPAM Polymer Gel Dosimeters on Three Orthogonal Planes and Temporal Stability Analysis.

Polymer gel dosimeters have been proven useful for dose evaluation in radiotherapy treatments. Previous studies have demonstrated that using a polymer gel dosimeter requires a 24 h reaction time to stabilize and further evaluate the measured dose distribution in two-dimensional dosimetry. In this study, the short-term stability within 24 h and feasibility of N-isopropylacrylamide (NIPAM) polymer gel dosimeters for use in three-dimensional dosimetry were evaluated using magnetic resonance imaging (MRI). NIPAM gels were used to measure the dose volume in a clinical case of intensity-modulated radiation therapy (IMRT). For dose readouts, MR images of irradiated NIPAM gel phantoms were acquired at 2, 5, 12, and 24 h after dose delivery. The mean standard errors of dose conversion from using dose calibration curves (DRC) were calculated. The measured dose volumes at the four time points were compared with those calculated using a treatment planning system (TPS). The mean standard errors of the dose conversion from using the DRCs were lower than 1 Gy. Mean pass rates of 2, 5, 12, and 24 h axial dose maps calculated using gamma evaluation with 3% dose difference and 3 mm distance-to-agreement criteria were 83.5% ± 0.9%, 85.9% ± 0.6%, 98.7% ± 0.3%, and 98.5% ± 0.9%, respectively. Compared with the dose volume histogram of the TPS, the absolute mean relative volume differences of the 2, 5, 12, and 24 h measured dose volumes were lower than 1% for the irradiated region with an absorbed dose higher than 2.8 Gy. It was concluded that a 12 h reaction time was sufficient to acquire accurate dose volume using the NIPAM gels with MR readouts.

Small-Field Measurements of 3D Polymer Gel Dosimeters through Optical Computed Tomography.

With advances in therapeutic instruments and techniques, three-dimensional dose delivery has been widely used in radiotherapy. The verification of dose distribution in a small field becomes critical because of the obvious dose gradient within the field. The study investigates the dose distributions of various field sizes by using NIPAM polymer gel dosimeter. The dosimeter consists of 5% gelatin, 5% monomers, 3% cross linkers, and 5 mM THPC. After irradiation, a 24 to 96 hour delay was applied, and the gel dosimeters were read by a cone beam optical computed tomography (optical CT) scanner. The dose distributions measured by the NIPAM gel dosimeter were compared to the outputs of the treatment planning system using gamma evaluation. For the criteria of 3%/3 mm, the pass rates for 5 × 5, 3 × 3, 2 × 2, 1 × 1, and 0.5 × 0.5 cm2 were as high as 91.7%, 90.7%, 88.2%, 74.8%, and 37.3%, respectively. For the criteria of 5%/5 mm, the gamma pass rates of the 5 × 5, 3 × 3, and 2 × 2 cm2 fields were over 99%. The NIPAM gel dosimeter provides high chemical stability. With cone-beam optical CT readouts, the NIPAM polymer gel dosimeter has potential for clinical dose verification of small-field irradiation.

A novel method of estimating dose responses for polymer gels using texture analysis of scanning electron microscopy images.

Polymer gels are regarded as a potential dosimeter for independent validation of absorbed doses in clinical radiotherapy. Several imaging modalities have been used to convert radiation-induced polymerization to absorbed doses from a macro-scale viewpoint. This study developed a novel dose conversion mechanism by texture analysis of scanning electron microscopy (SEM) images. The modified N-isopropyl-acrylamide (NIPAM) gels were prepared under normoxic conditions, and were administered radiation doses from 5 to 20 Gy. After freeze drying, the gel samples were sliced for SEM scanning with 50×, 500×, and 3500× magnifications. Four texture indices were calculated based on the gray level co-occurrence matrix (GLCM). The results showed that entropy and homogeneity were more suitable than contrast and energy as dose indices for higher linearity and sensitivity of the dose response curves. After parameter optimization, an R (2) value of 0.993 can be achieved for homogeneity using 500× magnified SEM images with 27 pixel offsets and no outlier exclusion. For dose verification, the percentage errors between the prescribed dose and the measured dose for 5, 10, 15, and 20 Gy were -7.60%, 5.80%, 2.53%, and -0.95%, respectively. We conclude that texture analysis can be applied to the SEM images of gel dosimeters to accurately convert micro-scale structural features to absorbed doses. The proposed method may extend the feasibility of applying gel dosimeters in the fields of diagnostic radiology and radiation protection.