ABSTRACT
Clinical measurements on breast cancer patients were performed with a three-dimensional tomographic photoacoustic prototype imager (PAM 2). Patients with a suspicious lesion, visiting the center for breast care of a local hospital, were included in the study. The acquired photoacoustic images were compared to conventional clinical images. Of 30 scanned patients, 19 were diagnosed with one or more malignancies, of which a subset of four patients was selected for detailed analysis. Reconstructed images were processed to enhance image quality and the visibility of blood vessels. Processed photoacoustic images were compared to contrast-enhanced magnetic resonance images where available, which aided in localizing the expected tumoral region. In two cases, spotty high-intensity photoacoustic signals could be seen in the tumoral region, attributable to the tumor. One of these cases also displayed a relatively high image entropy at the tumor site, likely related to the chaotic vascular networks associated with malignancies. For the other two cases, it was not possible to identify features indicative of malignancy, because of limitations in the illumination scheme and difficulties in locating the region of interest in the photoacoustic image.
Subject(s)
Breast Neoplasms , Humans , Female , Breast Neoplasms/diagnostic imaging , Diagnostic Imaging , Spectrum Analysis , Breast/diagnostic imaging , EntropyABSTRACT
In the last decade, multiparametric magnetic resonance imaging (mpMRI) has been expanding its role in prostate cancer detection and characterization. In this work, 19 patients with clinically significant peripheral zone (PZ) tumours were studied. Tumour masks annotated on the whole-mount histology sections were mapped on T2-weighted (T2w) and diffusion-weighted (DW) sequences. Gray-level histograms of tumoral and normal tissue were compared using six first-order texture features. Multivariate analysis of variance (MANOVA) was used to compare group means. Mean intensity signal of ADC showed the highest showed the highest area under the receiver operator characteristics curve (AUC) equal to 0.85. MANOVA analysis revealed that ADC features allows a better separation between normal and cancerous tissue with respect to T2w features (ADC: P = 0.0003, AUC = 0.86; T2w: P = 0.03, AUC = 0.74). MANOVA proved that the combination of T2-weighted and apparent diffusion coefficient (ADC) map features increased the AUC to 0.88. Histogram-based features extracted from invivo mpMRI can help discriminating significant PZ PCa.