Critique of optical coherence tomography in epistemological metrology

TL;DR

This paper proposes a philosophical framework called epistemological-metrology to analyze modern measurement systems like optical coherence tomography (OCT) with AI, emphasizing the measurement of concepts and understanding processes.

Contribution

It introduces a novel theoretical model for understanding modern measurement, especially in complex systems like OCT with AI, focusing on constructive and conceptual measurement processes.

Findings

Classifies measurement acts into sensing, understanding, and reasoning.

Applies the framework to analyze OCT-related measurements and AI-based diagnosis.

Highlights the constructive nature of understanding and reasoning measurements.

Abstract

This treaties aims to introduce a speculative framework to comprehend modern metrology. In a classical sense, measurement measures physically existing quantities, such as length, mass, and time. By contrast, modern measurement measures a wide spectrum of objects, for example, an imaging biomarker, which is a user-defined pattern. Artificial-intelligence (AI)-based diagnosis is another example. If we regard an optical coherence tomography (OCT) device equipped with diagnostic AI as a comprehensive "measurement system," the object being measured is a "disease," which is not a physical quantity but a concept. To comprehend the wide range of modern measurements, we introduce a speculative, i.e., philosophical and theoretical, model called the "epistemological-metrology model." In this model, we describe the act of measurement as cascading encoding-and-decoding processes. In the encoding processes, three types of objects-to-be measured are considered, which include substance, existence, and concept. Then we classify acts of measurement into "sensing," "understanding," and "reasoning," which measure the substance, existence, and concept, respectively. We note that the measurements in the understanding and reasoning classes are constructive. Namely, they proactively define the quantity-to-be-measured by the measurement modalities themselves. A speculative method to warrant the relevance of such constructive measurements is presented. We investigate several modern OCT-related measurements, including AI-based diagnosis, types of polarization sensitive OCT, and attenuation coefficient imaging, using our theoretical framework.