Towards in vivo characterization of thyroid nodules suspicious for malignancy using multispectral optoacoustic tomography.

PURPOSE: Patient-tailored management of thyroid nodules requires improved risk of malignancy stratification by accurate preoperative nodule assessment, aiming to personalize decisions concerning diagnostics and treatment. Here, we perform an exploratory pilot study to identify possible patterns on multispectral optoacoustic tomography (MSOT) for thyroid malignancy stratification. For the first time, we directly correlate MSOT images with histopathology data on a detailed level. METHODS: We use recently enhanced data processing and image reconstruction methods for MSOT to provide next-level image quality by means of improved spatial resolution and spectral contrast. We examine optoacoustic features in thyroid nodules associated with vascular patterns and correlate these directly with reference histopathology. RESULTS: Our methods show the ability to resolve blood vessels with diameters of 250 µm at depths of up to 2 cm. The vessel diameters derived on MSOT showed an excellent correlation (R2-score of 0.9426) with the vessel diameters on histopathology. Subsequently, we identify features of malignancy observable in MSOT, such as intranodular microvascularity and extrathyroidal extension verified by histopathology. Despite these promising features in selected patients, we could not determine statistically relevant differences between benign and malignant thyroid nodules based on mean oxygen saturation in thyroid nodules. Thus, we illustrate general imaging artifacts of the whole field of optoacoustic imaging that reduce image fidelity and distort spectral contrast, which impedes quantification of chromophore presence based on mean concentrations. CONCLUSION: We recommend examining optoacoustic features in addition to chromophore quantification to rank malignancy risk. We present optoacoustic images of thyroid nodules with the highest spatial resolution and spectral contrast to date, directly correlated to histopathology, pushing the clinical translation of MSOT.

Image processing improvements afford second-generation handheld optoacoustic imaging of breast cancer patients.

Background: Since the initial breast transillumination almost a century ago, breast cancer imaging using light has been considered in different implementations aiming to improve diagnostics, minimize the number of available biopsies, or monitor treatment. However, due to strong photon scattering, conventional optical imaging yields low resolution images, challenging quantification and interpretation. Optoacoustic imaging addresses the scattering limitation and yields high-resolution visualization of optical contrast, offering great potential value for breast cancer imaging. Nevertheless, the image quality of experimental systems remains limited due to a number of factors, including signal attenuation with depth and partial view angle and motion effects, particularly in multi-wavelength measurements. Methods: We developed data analytics methods to improve the accuracy of handheld optoacoustic breast cancer imaging, yielding second-generation optoacoustic imaging performance operating in tandem with ultrasonography. Results: We produced the most advanced images yet with handheld optoacoustic examinations of the human breast and breast cancer, in terms of resolution and contrast. Using these advances, we examined optoacoustic markers of malignancy, including vasculature abnormalities, hypoxia, and inflammation, on images obtained from breast cancer patients. Conclusions: We achieved a new level of quality for optoacoustic images from a handheld examination of the human breast, advancing the diagnostic and theranostic potential of the hybrid optoacoustic-ultrasound (OPUS) examination over routine ultrasonography.

VerSe: A Vertebrae labelling and segmentation benchmark for multi-detector CT images.

Vertebral labelling and segmentation are two fundamental tasks in an automated spine processing pipeline. Reliable and accurate processing of spine images is expected to benefit clinical decision support systems for diagnosis, surgery planning, and population-based analysis of spine and bone health. However, designing automated algorithms for spine processing is challenging predominantly due to considerable variations in anatomy and acquisition protocols and due to a severe shortage of publicly available data. Addressing these limitations, the Large Scale Vertebrae Segmentation Challenge (VerSe) was organised in conjunction with the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) in 2019 and 2020, with a call for algorithms tackling the labelling and segmentation of vertebrae. Two datasets containing a total of 374 multi-detector CT scans from 355 patients were prepared and 4505 vertebrae have individually been annotated at voxel level by a human-machine hybrid algorithm (https://osf.io/nqjyw/, https://osf.io/t98fz/). A total of 25 algorithms were benchmarked on these datasets. In this work, we present the results of this evaluation and further investigate the performance variation at the vertebra level, scan level, and different fields of view. We also evaluate the generalisability of the approaches to an implicit domain shift in data by evaluating the top-performing algorithms of one challenge iteration on data from the other iteration. The principal takeaway from VerSe: the performance of an algorithm in labelling and segmenting a spine scan hinges on its ability to correctly identify vertebrae in cases of rare anatomical variations. The VerSe content and code can be accessed at: https://github.com/anjany/verse.