An algorithm for PET tumor volume and activity quantification: without specifying camera's point spread function (PSF).

PURPOSE: The authors have developed an algorithm for segmentation and removal of the partial volume effect (PVE) of tumors in positron emission tomography (PET) images. The algorithm accurately measures functional volume (FV) and activity concentration (AC) of tumors independent of the camera's full width half maximum (FWHM). METHODS: A novel iterative histogram thresholding (HT) algorithm is developed to segment the tumors in PET images, which have low resolution and suffer from inherent noise in the image. The algorithm is initiated by manually drawing a region of interest (ROI). The segmented tumors are subjected to the iterative deconvolution thresholding segmentation (IDTS) algorithm, where the Van-Cittert's method of deconvolution is used for correcting PVE. The IDTS algorithm is fully automated and accurately measures the FV and AC, and stops once it reaches convergence. The convergence criteria or stopping conditions are developed in such a way that the algorithm does not rely on estimating the FWHM of the point spread function (PSF) to perform the deconvolution process. The algorithm described here was tested in phantom studies, where hollow spheres (0.5-16 ml) were used to represent tumors with a homogeneous activity distribution, and an irregular shaped volume was used to represent a tumor with a heterogeneous activity distribution. The phantom studies were performed with different signal to background ratios (SBR) and with different acquisition times (1 min, 3 min, and 5 min). The parameters in the algorithm were also changed (FWHM and matrix size of the Gaussian function) to check the accuracy of the algorithm. Simulated data were also used to test the algorithm with tumors having heterogeneous activity distribution. RESULTS: The results show that changing the size and shape of the ROI during initiation of the algorithm had no significant impact on the FV. An average FV overestimation of 30% and an average AC underestimation of 35% were observed for the smallest tumor (0.5 ml) over the entire range of noise and SBR level. The difference in average FV and AC estimations from the actual volumes were less than 5% as the tumor size increased to 16 ml. For tumors with heterogeneous activity profile, the overall volume error was less than 10%. The average overestimation of FV was less than 10% and classification error was around 11%. CONCLUSIONS: The algorithm developed herein was extensively tested and is not dependent on accurately quantifying the camera's PSF. This feature demonstrates the robustness of the algorithm and enables it to be applied on a wide range of noise and SBR within an image. The ultimate goal of the algorithm is to be able to be operated independent of the camera type used and the reconstruction algorithm deployed.

High-plex imaging of RNA and proteins at subcellular resolution in fixed tissue by spatial molecular imaging.

Resolving the spatial distribution of RNA and protein in tissues at subcellular resolution is a challenge in the field of spatial biology. We describe spatial molecular imaging, a system that measures RNAs and proteins in intact biological samples at subcellular resolution by performing multiple cycles of nucleic acid hybridization of fluorescent molecular barcodes. We demonstrate that spatial molecular imaging has high sensitivity (one or two copies per cell) and very low error rate (0.0092 false calls per cell) and background (~0.04 counts per cell). The imaging system generates three-dimensional, super-resolution localization of analytes at ~2 million cells per sample. Cell segmentation is morphology based using antibodies, compatible with formalin-fixed, paraffin-embedded samples. We measured multiomic data (980 RNAs and 108 proteins) at subcellular resolution in formalin-fixed, paraffin-embedded tissues (nonsmall cell lung and breast cancer) and identified >18 distinct cell types, ten unique tumor microenvironments and 100 pairwise ligand-receptor interactions. Data on >800,000 single cells and ~260 million transcripts can be accessed at http://nanostring.com/CosMx-dataset .

A low-interaction automatic 3D liver segmentation method using computed tomography for selective internal radiation therapy.

This study introduces a novel liver segmentation approach for estimating anatomic liver volumes towards selective internal radiation treatment (SIRT). The algorithm requires minimal human interaction since the initialization process to segment the entire liver in 3D relied on a single computed tomography (CT) slice. The algorithm integrates a localized contouring algorithm with a modified k-means method. The modified k-means segments each slice into five distinct regions belonging to different structures. The liver region is further segmented using localized contouring. The novelty of the algorithm is in the design of the initialization masks for region contouring to minimize human intervention. Intensity based region growing together with novel volume of interest (VOI) based corrections is used to accomplish the single slice initialization. The performance of the algorithm is evaluated using 34 liver CT scans. Statistical experiments were performed to determine consistency of segmentation and to assess user dependency on the initialization process. Volume estimations are compared to the manual gold standard. Results show an average accuracy of 97.22% for volumetric calculation with an average Dice coefficient of 0.92. Statistical tests show that the algorithm is highly consistent (P = 0.55) and independent of user initialization (P = 0.20 and Fleiss' Kappa = 0.77 ± 0.06).

A 3-D liver segmentation method with parallel computing for selective internal radiation therapy.

This study describes a new 3-D liver segmentation method in support of the selective internal radiation treatment as a treatment for liver tumors. This 3-D segmentation is based on coupling a modified k-means segmentation method with a special localized contouring algorithm. In the segmentation process, five separate regions are identified on the computerized tomography image frames. The merit of the proposed method lays in its potential to provide fast and accurate liver segmentation and 3-D rendering as well as in delineating tumor region(s), all with minimal user interaction. Leveraging of multicore platforms is shown to speed up the processing of medical images considerably, making this method more suitable in clinical settings. Experiments were performed to assess the effect of parallelization using up to 442 slices. Empirical results, using a single workstation, show a reduction in processing time from 4.5 h to almost 1 h for a 78% gain. Most important is the accuracy achieved in estimating the volumes of the liver and tumor region(s), yielding an average error of less than 2% in volume estimation over volumes generated on the basis of the current manually guided segmentation processes. Results were assessed using the analysis of variance statistical analysis.