First-in-Human (FiH) Clinical Investigation
The Wavelia First-in-Human clinical investigation took place at NUIG Clinical Research Facility Galway, in Ireland. For this clinical investigation, a cohort of 25 patients with 3 different types of pathologies (invasive carcinomas, biopsy-confirmed benign lesions, simple cysts) were enrolled.
The primary objectives of this study were to collect data to evaluate the safety of Wavelia and to refine the imaging system towards development of a novel, adjunct, breast imaging modality. This study also allowed to explore the capacity of Wavelia to detect and characterize pre-diagnosed breast pathologies.
Main study outcomes:
Breast lesion detectability:
Correlation of the image with the breast pathology was achieved for 21 out of the 24 patients considered in the analysis, based on findings from conventional imaging, as well as surgical findings for the cancerous lesions. Nine (9) out of the 11 cancers and 12 out of the 13 benign breast pathologies were detectable with Wavelia. The global lesion detectability rate was 87.5%, with 75% of the breast lesions being detected using a fixed setting of the imaging algorithm. The two cancers which were not detectable were smaller than 10mm.
Breast lesion sizing:
For the cancer patients, the post-surgery histological total size of the excised tumor was used as gold standard. The Wavelia results for the tumor size showed positive correlation with the surgical findings, and a preliminary tendency to perform better than conventional imaging for estimating the size of the large and partially visible Invasive Lobular Carcinomas. This promising outcome remains to be confirmed with larger sample sizes, in the next clinical investigations.
Discrimination between malignant and benign breast lesions:
In this analysis, features related to the shape and texture of the breast lesion were computed on Regions-of-Interest which were automatically extracted from the images and associated to the detection of the breast lesion. An interesting potential for discrimination between malignant and benign breast lesions using 3 features was demonstrated in the study. Two classifiers were trained on the small available patient dataset and showed a rate of mis-classification of 11.5% only. This is a result with confirmed clinical interest which is sought to be reproduced on larger and more diverse patient datasets in the next clinical investigations.
