Diagnosis and Localization of Prostate Cancer via Automated Multiparametric MRI Equipped with Artificial Intelligence

Prostate MRI scans for pre-biopsied patients are important. However, fewer radiologists are available for MRI diagnoses, which requires multi-sequential interpretations of multi-slice images. To reduce such a burden, artificial intelligence (AI)-based, computer-aided diagnosis is expected to be a critical technology. We present an AI-based method for pinpointing prostate cancer location and determining tumor morphology using multiparametric MRI. The study enrolled 15 patients who underwent radical prostatectomy between April 2008 and August 2017 at our institution. We labeled the cancer area on the peripheral zone on MR images, comparing MRI with histopathological mapping of radical prostatectomy specimens. Likelihood maps were drawn, and tumors were divided into morphologically distinct regions using the superpixel method. Likelihood maps consisted of pixels, which utilize the cancer likelihood value computed from the T2-weighted, apparent diffusion coefficient, and diffusion-weighted MRI-based texture features. Cancer location was determined based on the likelihood maps. We evaluated the diagnostic performance by the area under the receiver operating characteristic (ROC) curve according to the Chi-square test. The area under the ROC curve was 0.985. Sensitivity and specificity for our approach were 0.875 and 0.961 (p < 0.01), respectively. Our AI-based procedures were successfully applied to automated prostate cancer localization and shape estimation using multiparametric MRI.

A Nonlinear Data-Driven Towed Array Shape Estimation Method Using Passive Underwater Acoustic Data

Large-aperture towed linear hydrophone array has been widely used for beamforming-based signal enhancement in passive sonar systems; however, its performance can drastically decrease due to the array distortion caused by rapid tactical maneuvers of the towed platform, oceanic currents, hydrodynamic effects, etc. In this paper, an enhanced data-driven shape array estimation scheme is provided in the passive underwater acoustic data, and a novel nonlinear outlier-robust particle filter (ORPF) method is proposed to acquire enhanced estimates of time delays in the presence of distorted hydrophone array. A conventional beamforming technique based on a hypothetical array is first used, and the detection of the narrow-band components is sequentially carried out so that the corresponding amplitudes and phases at these narrow-band components can be acquired. We convert the towed array estimation problem into a nonlinear discrete-time filtering problem with the joint estimates of amplitudes and time-delay differences, and then propose the ORPF method to acquire enhanced estimates of the time delays by exploiting the underlying properties of slowly changing time-delay differences across sensors. The proposed scheme fully exploits directional radiated noise targets as sources of opportunity for online array shape estimation, and thus it requires neither the number nor direction of sources to be known in advance. Both simulations and real experimental data show the effectiveness of the proposed method.

A Comparative Evaluation of Snowflake Particle Size and Shape Estimation Techniques used by the Precipitation Imaging Package (PIP), Multi-Angle Snowflake Camera (MASC), and Two-Dimensional Video Disdrometer (2DVD)

Measurements of snowflake particle size and shape are important for studying the snow microphysics. While a number of instruments exist that are designed to measure these important parameters, this study focuses on the measurement techniques of three digital video disdrometers: the Precipitation Imaging Package (PIP), the Multi-Angle Snowflake Camera (MASC) and the Two-Dimensional Video Disdrometer (2DVD). To gain a better understanding of the relative strengths and weaknesses of these instruments and to provide a foundation upon which comparisons can be made between studies using data from different instruments, we perform a comparative analysis of the measurement algorithms employed by each of the three instruments by applying the algorithms to snowflake images captured by PIP during the ICEP-POP 2018 field campaign. Our analysis primarily focuses on the measurements of area, equivalent diameter, and aspect ratio. Our findings indicate that area and equi-area diameter measurements using the 2DVD camera setup should be the most accurate, followed by MASC, which is slightly more accurate than PIP. In terms of the precision of the area and equi-area diameter measurements, however, MASC is considerably more precise than PIP or 2DVD, which provide similar precision once the effects of the PIP image compression algorithm are taken into account. Both PIP and MASC use shape-fitting algorithms to measure aspect ratio. While our analysis of the MASC aspect ratio suggests the measurements are reliable, our findings indicate that both the ellipse and rectangle aspect ratios produced by PIP under-performed considerably due to the shortcomings of the PIP shape-fitting techniques. That said, we also demonstrate that reliable measurements of aspect ratio can be retrieved from PIP by reprocessing the PIP images using either the MASC shape-fitting technique or a tensor-based ellipse-fitting technique. Because of differences in instrument design, 2DVD produces measurements of particle horizontal and vertical extent rather than length and width. Furthermore, the 2DVD measurements of particle horizontal extent can be contaminated by horizontal particle motion. Our findings indicate that, although the correction technique used to remove the horizontal motion contamination performs remarkably well with snowflakes despite being designed for use with rain drops, the 2DVD measurements of particle horizontal extent are potentially unreliable.

Kimera: From SLAM to spatial perception with 3D dynamic scene graphs

Humans are able to form a complex mental model of the environment they move in. This mental model captures geometric and semantic aspects of the scene, describes the environment at multiple levels of abstractions (e.g., objects, rooms, buildings), includes static and dynamic entities and their relations (e.g., a person is in a room at a given time). In contrast, current robots’ internal representations still provide a partial and fragmented understanding of the environment, either in the form of a sparse or dense set of geometric primitives (e.g., points, lines, planes, and voxels), or as a collection of objects. This article attempts to reduce the gap between robot and human perception by introducing a novel representation, a 3D dynamic scene graph (DSG), that seamlessly captures metric and semantic aspects of a dynamic environment. A DSG is a layered graph where nodes represent spatial concepts at different levels of abstraction, and edges represent spatiotemporal relations among nodes. Our second contribution is Kimera, the first fully automatic method to build a DSG from visual–inertial data. Kimera includes accurate algorithms for visual–inertial simultaneous localization and mapping (SLAM), metric–semantic 3D reconstruction, object localization, human pose and shape estimation, and scene parsing. Our third contribution is a comprehensive evaluation of Kimera in real-life datasets and photo-realistic simulations, including a newly released dataset, uHumans2, which simulates a collection of crowded indoor and outdoor scenes. Our evaluation shows that Kimera achieves competitive performance in visual–inertial SLAM, estimates an accurate 3D metric–semantic mesh model in real-time, and builds a DSG of a complex indoor environment with tens of objects and humans in minutes. Our final contribution is to showcase how to use a DSG for real-time hierarchical semantic path-planning. The core modules in Kimera have been released open source.