Where respiratory imaging becomes precise, quantified volumetric insight.
CORE Approach
BreathIn is a medical imaging analysis pipeline designed to quantify lung breathing volumes and measure chest wall dynamics across respiratory phases. It processes raw CT scans from multiple patients through automated segmentation, volumetric computation, and bony landmark measurement — all in one streamlined workflow.
The Core Intent of Our Approach
Our approach focused on transforming compressed DICOM CT data into clinically meaningful measurements while ensuring accuracy, reproducibility, and clear visualization across all patients and respiratory phases.
Quantify breathing volumes
egment right and left lungs across inspiration, expiration, and resting phases to calculate precise air volumes in milliliters.
Measure chest wall dynamics
Extract anterior–posterior and lateral dimensions at bony landmarks to capture how the thoracic cavity changes with each breath.
Automate the pipeline
Replace manual measurement with TotalSegmentator-driven segmentation, batch processing, and structured CSV output for every patient and phase.
Visualize with clarity
Generate high-quality 2D plots, 3D interactive models, and mid-sagittal overlays that make volumetric differences immediately interpretable.
THE PROBLEM
Traditional chest CT analysis workflows are manual, fragmented, and inconsistent — creating bottlenecks for clinicians and researchers trying to quantify breathing mechanics.
Compressed raw data
DICOM CT files arrive with lossless JPEG compression, requiring decompression and format conversion before any analysis can begin.
Manual segmentation
Identifying lung boundaries and bony landmarks by hand is time-consuming, subjective, and prone to inter-observer variability.
Lack of standardization
No consistent methodology for measuring chest wall dimensions across patients, phases, and anatomical levels.
Processing constraints
Large scan sizes push memory and compute limits, slowing batch analysis across multiple patients and respiratory phases.
WORKFLOW
A structured, step-by-step approach that transformed raw compressed CT data into precise volumetric measurements and reproducible chest wall analysis.
Prepare
data
We decompressed DICOM files, converted to NIfTI format, and organized patient data by respiratory phase into a clean directory structure.
Segment
anatomy
TotalSegmentator generated masks for lungs, ribs, sternum, and costal cartilage — automating what was previously done by hand.
Extract
measurements
Air volumes were calculated via HU thresholding, and chest dimensions were measured at bony landmarks using distance-based localization.
Deliver
results
Structured CSVs, mid-sagittal plots, axial overlays, and interactive 3D models were generated for every patient and phase.
TECH INSIGHTS
The technical foundation that enables automated lung segmentation, precise bony landmark measurement, and reproducible volumetric analysis across patients and respiratory phases.
Total
Segmentator
Automated deep learning–based segmentation of lungs, ribs, sternum, and costal cartilage from NIfTI-converted CT scans.
HU
thresholding
Voxels below −400 Hounsfield Units are classified as air-filled regions, enabling accurate lung air volume calculation in milliliters.
Distance-based mapping between ribs, costal cartilage, and sternum to identify sternocostal junctions at the 2nd and 3rd rib levels.
GPU-accelerated
Optimized scripts handle large scan sizes efficiently, overcoming memory constraints during batch analysis across all patients and phases.
Structured
outputs
Results compiled into standardized CSVs with subject ID, phase, lung volumes, landmark dimensions, and lateral widths for easy downstream use.
3D
visualization
Interactive HTML-based 3D models and mid-sagittal overlays allow clinicians to explore volumetric and dimensional changes from any angle.
Result
BreathIn transformed raw compressed CT data into precise, reproducible respiratory measurements — giving clinicians clear volumetric insight across every patient and breathing phase.
Automated segmentation
|
Precise measurement
|
Reproducible results
|
Clinical clarity
CASE STUDIES
Explore related projects with similar challenges and outcomes.