Value of contrast-enhanced CT based radiomic machine learning algorithm in differentiating gastrointestinal stromal tumors with KIT exon 11 mutation: a two-center study.

PURPOSE Knowing the genetic phenotype of gastrointestinal stromal tumors (GISTs) is essential for patients who receive therapy with tyrosine kinase inhibitors. The aim of this study was to develop a radiomic algorithm for predicting GISTs with KIT exon 11 mutation. METHODS We enrolled 106 patients (80 in the training set, 26 in the validation set) with clinicopathologically confirmed GISTs from two centers. Preoperative and postoperative clinical characteristics were selected and analyzed to construct the clinical model. Arterial phase, venous phase, delayed phase, and tri-phase combined radiomics algorithms were generated from the training set based on contrast-enhanced computed tomography (CE-CT) images. Various radiomics feature selection methods were used, namely least absolute shrinkage and selection operator (LASSO); minimum redundancy maximum relevance (mRMR); and generalized linear model (GLM) as a machine-learning classifier. Independent predictive factors were determined to construct preoperative and postoperative radiomics nomograms by multivariate logistic regression analysis. The performances of the clinical model, radiomics algorithm, and radiomics nomogram in distinguishing GISTs with the KIT exon 11 mutation were evaluated by area under the curve (AUC) of the receiver operating characteristics. RESULTS Of 106 patients who underwent genetic analysis, 61 had the KIT exon 11 mutation. The combined radiomics algorithm was found to be the best prediction model for differentiating the expression status of the KIT exon 11 mutation (AUC = 0.836; 95% confidence interval [CI], 0.640-0.951) in the validation set. The clinical model, and preoperative and postoperative radiomics nomograms had AUCs of 0.606 (95% CI, 0.397-0.790), 0.715 (95% CI, 0.506-0.873), and 0.679 (95% CI, 0.468-0.847), respectively, with the validation set. CONCLUSION The radiomics algorithm could distinguish GISTs with the KIT exon 11 mutation based on CE-CT images and could potentially be used for selective genetic analysis to support the precision medicine of GISTs.

A high-frequency linear ultrasonic array utilizing an interdigitally bonded 2-2 piezo-composite.

This paper describes the development of a high-frequency 256-element linear ultrasonic array utilizing an interdigitally bonded (IB) piezo-composite. Several IB composites were fabricated with different commercial and experimental piezoelectric ceramics and evaluated to determine a suitable formulation for use in high-frequency linear arrays. It was found that the fabricated fine-scale 2-2 IB composites outperformed 1-3 IB composites with identical pillar- and kerf-widths. This result was not expected and lead to the conclusion that dicing damage was likely the cause of the discrepancy. Ultimately, a 2-2 composite fabricated using a fine-grain piezoelectric ceramic was chosen for the array. The composite was manufactured using one IB operation in the azimuth direction to produce approximately 19-µm-wide pillars separated by 6-µm-wide kerfs. The array had a 50 µm (one wavelength in water) azimuth pitch, two matching layers, and 2 mm elevation length focused to 7.3 mm using a polymethylpentene (TPX) lens. The measured pulse-echo center frequency for a representative array element was 28 MHz and -6-dB bandwidth was 61%. The measured single-element transmit -6-dB directivity was estimated to be 50°. The measured insertion loss was 19 dB after compensating for the effects of attenuation and diffraction in the water bath. A fine-wire phantom was used to assess the lateral and axial resolution of the array when paired with a prototype system utilizing a 64-channel analog beamformer. The -6-dB lateral and axial resolutions were estimated to be 125 and 68 µm, respectively. An anechoic cyst phantom was also imaged to determine the minimum detectable spherical inclusion, and thus the 3-D resolution of the array and beamformer. The minimum anechoic cyst detected was approximately 300 µm in diameter.

Development of a 64 channel ultrasonic high frequency linear array imaging system.

In order to improve the lateral resolution and extend the field of view of a previously reported 48 element 30 MHz ultrasound linear array and 16-channel digital imaging system, the development of a 256 element 30 MHz linear array and an ultrasound imaging system with increased channel count has been undertaken. This paper reports the design and testing of a 64 channel digital imaging system which consists of an analog front-end pulser/receiver, 64 channels of Time-Gain Compensation (TGC), 64 channels of high-speed digitizer as well as a beamformer. A Personal Computer (PC) is used as the user interface to display real-time images. This system is designed as a platform for the purpose of testing the performance of high frequency linear arrays that have been developed in house. Therefore conventional approaches were taken it its implementation. Flexibility and ease of use are of primary concern whereas consideration of cost-effectiveness and novelty in design are only secondary. Even so, there are many issues at higher frequencies but do not exist at lower frequencies need to be solved. The system provides 64 channels of excitation pulsers while receiving simultaneously at a 20-120 MHz sampling rate to 12-bits. The digitized data from all channels are first fed through Field Programmable Gate Arrays (FPGAs), and then stored in memories. These raw data are accessed by the beamforming processor to re-build the image or to be downloaded to the PC for further processing. The beamformer that applies delays to the echoes of each channel is implemented with the strategy that combines coarse (8.3 ns) and fine delays (2 ns). The coarse delays are integer multiples of the sampling clock rate and are achieved by controlling the write enable pin of the First-In-First-Out (FIFO) memory to obtain valid beamforming data. The fine delays are accomplished with interpolation filters. This system is capable of achieving a maximum frame rate of 50 frames per second. Wire phantom images acquired with this system show a spatial resolution of 146 µm (lateral) and 54 µm (axial). Images with excised rabbit and pig eyeball as well as mouse embryo were also acquired to demonstrate its imaging capability.

A high-frequency, high frame rate duplex ultrasound linear array imaging system for small animal imaging.

High-frequency (HF) ultrasound imaging has been shown to be useful for non-invasively imaging anatomical structures of the eye and small animals in biological and pharmaceutical research, achieving superior spatial resolution. Cardiovascular research utilizing mice requires not only realtime B-scan imaging, but also ultrasound Doppler to evaluate both anatomy and blood flow of the mouse heart. This paper reports the development of an HF ultrasound duplex imaging system capable of both B-mode imaging and Doppler flow measurements, using a 64-element linear array. The system included an HF pulsed-wave Doppler module, a 32-channel HF B-mode imaging module, a PC with a 200 MS/s 14-bit A/D card, and real-time LabView software. A 50 dB SNR and a depth of penetration of larger than 12 mm were achieved using a 35-MHz linear array with 50 µm pitch. The two-way beam widths were determined to be 165 to 260 µm and the clutter-energy-to-total-energy ratio (CTR) were 9.1 to 12 dB when the array was electronically focused at different focal points at depths from 4.8 to 9.6 mm. The system is capable of acquiring real-time B-mode images at a rate greater than 400 frames per second (fps) for a 4.8 x 13 mm field of view, using a 30- MHz 64-element linear array with 100 µm pitch. Sample in vivo cardiac high frame rate images and duplex images of mouse hearts are shown to assess its current imaging capability and performance for small animals.

Real-time assessment of flow reversal in an eccentric arterial stenotic model.

Plaque rupture is the leading cause of acute coronary syndromes and stroke. Plaque formation, otherwise known as stenosis, preferentially occurs in the regions of arterial bifurcation or curvatures. To date, real-time assessment of stenosis-induced flow reversal remains a clinical challenge. By interfacing microelectromechanical system (MEMS) thermal sensors with the high frequency pulsed wave (PW) Doppler ultrasound, we proposed to assess flow reversal in the presence of an eccentric stenosis. We developed a 3-D stenotic model (inner diameter of 6mm, an eccentric stenosis with a height of 2.75 mm, and width of 21 mm) simulating a superficial arterial vessel. We demonstrated that heat transfer from the sensing element (2 x 80 µm²) to the flow field peaked as a function of flow rates at the throat of the stenosis along the center/midline of arterial model, and dropped downstream from the stenosis, where flow reversal was detected by the high frequency ultrasound device at 45 MHz. Computational fluid dynamics (CFD) codes are in agreement with the ultrasound-acquired flow profiles upstream, downstream, and at the throat of the stenosis. Hence, we characterized regions of eccentric stenosis in terms of changes in heat transfer along the midline of vessel and identified points of flow reversal with high spatial and temporal resolution.