Luis Carvalho

Title: Luis Carvalho and Machine-Learning Methods for Immunostain Analysis.

Introduction

Luis Carvalho is an inventor based in Valencia, California. His patent work addresses the challenge of training machine-learning systems with noisy or inaccurate immunostain annotations, with an approach designed to refine virtually stained tissue images.

Latest Patents

Carvalho is named on one patent, 1 patent: “Systems and Methods for Training Machine Learning on Noisy and Inaccurate Immunostains.” The invention describes a method for refining annotations in virtually stained images. It separates a stained tissue image into counterstain and immunostaining signals using a linear transformation, then identifies and segments nuclei to create single-cell regions.

The method compares immunostaining signals within those regions against a user-defined threshold to identify true-positive cells. It then refines the immunostaining image using information about protein localization or tissue morphology, together with the identified true-positive cells. Finally, it overlays the counterstain image and refined immunostaining signals to produce a refined virtually stained image annotation.

Career Highlights

Carvalho is associated with Pictor Labs, Inc. His patent reflects work at the intersection of machine learning and tissue-image analysis, particularly in improving the reliability of annotations derived from immunostaining.

Collaborations

Hirofumi Kobayashi and Yair Rivenson are identified as Carvalho’s coworkers. Their shared association with Pictor Labs places Carvalho’s invention within a collaborative environment focused on computational approaches to biological imaging.

Conclusion

Luis Carvalho’s patent presents a structured approach to refining virtually stained image annotations by combining cell segmentation, threshold-based identification, and tissue-specific information.