L.I.V.E. — Lung Internal Visualization Engine

The Problem

Respiratory disease is diagnosed largely through manual auscultation — a clinician listening with a stethoscope and relying on subjective judgment and experience. This makes lung sound assessment inconsistent between providers, hard to document over time, and difficult to monitor remotely.

There is no standardized, repeatable way to capture lung sounds across the whole back and compare them over time or between patients. A physician typically listens at a handful of points, for a few seconds each, and records only written impressions — not the underlying audio. That leaves no objective record to revisit, share with a specialist, or run through automated analysis. L.I.V.E. is motivated by closing that gap: giving clinicians a consistent, automated, and data-driven way to capture and interpret lung sounds.

Goals & Approach

L.I.V.E. is a robotic system that automates auscultation: it drives a stethoscope to a series of defined points across the back, presses it against the body, and records lung sounds at each location for a standardized 20 seconds. Those recordings are streamed through a Raspberry Pi, compared against a database of known lung sounds, and turned into a visual map of lung health — a diagnostic aid for the physician, not the patient.

Design and build the robot chassis and mechanism that moves the stethoscope across the body and presses it firmly against the back at each recording point.
Mount and connect a stethoscope to a Raspberry Pi to capture clean audio at each point.
Record 20 seconds of audio at each specified point on the body, covering all major lung fields.
Build a database of reference lung sounds (normal and abnormal) to compare recordings against and detect anomalies.
Generate a 2D, color-coded visualization of the lungs from the recorded data, highlighting the location and likely type of any detected issue.
Work toward a 3D visualization as a stretch goal, giving physicians a fuller spatial map of lung condition.

Features & Functionality

Robotic Positioning

A robot moves the stethoscope across the body and presses it against the back at each specified auscultation point, automating what a clinician would otherwise do by hand.

Multi-Point Recording

Capture 20 seconds of lung sound audio at each back location through a stethoscope connected to a Raspberry Pi, building a complete acoustic picture of the lungs.

Lung Sound Mapping

Visualize recordings as a color-coded 2D map overlaying the lung structure, with a 3D visualization planned as a longer-term goal.

Abnormality Detection

Compare recordings against a database of known lung sounds to automatically flag wheezes, crackles, and other abnormal patterns.

Localized Diagnosis

Identify not just that an issue exists, but which part of the lung it's coming from — giving physicians a targeted starting point for further evaluation.

Current Status

L.I.V.E. is being built across two semesters of senior design, each broken into a planning checkpoint and a set of milestones with their own deliverables.

First Semester

Plan (Aug 31)	Plan, Presentation
Milestone 1 (Sep 28)	Requirement, Design, Test, Presentation, Progress Evaluation
Milestone 2 (Oct 26)	Presentation, Progress Evaluation
Milestone 3 (Nov 23)	Presentation, Progress Evaluation

Second Semester

Plan (?)	Plan, Presentation
Milestone 4 (?)	Presentation, Progress Evaluation
Milestone 5 (?)	Poster, Presentation, Progress Evaluation
Milestone 6 (?)	User and/or Developer Manual, Demo Video, Presentation, Progress Evaluation