Clarion Sentinel Platform · Pulmonary Embolism Division

Engine Technical
Design Document

Architecture, pipeline design, model specification, and performance validation across seven AI detection engines for pulmonary embolism intelligence.

Document Class
Technical Design Specification
Platform
Sentinel PE · Embolism Intelligence
Version
4.1.0
Classification
Confidential — Internal
Table of Contents
01CTPA Detection IntelligenceDeep learning clot detection on imaging02Clot Burden QuantificationAutomated Qanadli scoring & segmentation03Right Ventricular Strain AnalysisAutomated RV/LV ratio & cardiac assessment04Risk Stratification EngineMulti-modal severity classification05Pre-Test Probability IntelligenceWells/Geneva scoring & D-dimer integration06Hemodynamic PredictionDecompensation forecasting & shock detection07Treatment Response IntelligenceAnticoagulation monitoring & outcome tracking
Executive Summary

Pulmonary embolism is the third most fatal cardiovascular disease, with mortality exceeding 30% in massive PE when diagnosis or treatment is delayed. CTPA is the gold-standard diagnostic modality, but interpretation is time-consuming, susceptible to inter-reader variability, and frequently misses subsegmental emboli. Studies demonstrate that AI-powered CTPA analysis can complete comprehensive assessments within seconds — compared to average radiologist interpretation times of 40 minutes — while matching or exceeding human diagnostic accuracy.

Sentinel PE deploys seven AI detection engines spanning the full PE clinical pathway: from initial imaging detection and clot burden quantification, through automated right ventricular strain assessment and multi-modal risk stratification, to hemodynamic decompensation prediction and treatment response monitoring. Each engine integrates into existing PACS and EHR workflows via SMART on FHIR, enabling near-real-time notification to radiologists, emergency physicians, and pulmonary embolism response teams.

Model architectures include convolutional neural networks and transformer-based models for volumetric CT analysis, U-Net architectures for pixel-wise clot segmentation with Dice scores approaching 0.95, and attention-based deep learning frameworks achieving AUC of 0.95 for PE detection. A 2026 multicenter validation demonstrated superior diagnostic performance for combined manual + AI approaches (AUC 0.928–0.934) compared to either modality alone.

The platform directly addresses the core clinical challenge articulated by interventional cardiologists: the true killer in PE patients is failure of the right heart. Sentinel PE integrates both clot detection and cardiac strain assessment into a unified intelligence layer — enabling clinicians to triage patients and initiate treatment within minutes rather than hours of CTPA acquisition.

7
Analysis Engines
0.95
Detection AUC
<90s
Full Analysis
FDA
Clearance Path
Engine 01 · Imaging Intelligence Layer

CTPA Detection Intelligence

Every minute between scan acquisition and diagnosis is a minute the right ventricle may be failing. This engine eliminates that delay.

0.95
AUC
1.3s
Inference
94.8%
Sensitivity
Processing Pipeline
01
DICOM Ingestion
Real-time CTPA series capture from PACS. Automated protocol validation: contrast timing, slice thickness (0.625mm), kV verification. Quality gating for motion artifacts.
DICOMPACSQA Gate
02
Vessel Segmentation
3D pulmonary arterial tree extraction via U-Net. Main, lobar, segmental, and subsegmental branches mapped to anatomical atlas with vessel-level labeling.
U-Net3D Seg.Atlas
03
Clot Detection
Attention-based CNN–LSTM network scans volumetric data for filling defects. Trained on 7,279 CTPA studies (RSPECT dataset) with multi-center external validation.
CNN-LSTMAttention
04
Location Classification
Emboli classified by anatomical position: central (saddle), main PA, lobar, segmental, subsegmental. Each tagged with confidence and laterality.
Anatomical MapLaterality
05
Triage Alert
PACS worklist reprioritization for positive cases. Push notification to ED, radiology, and PERT. Downstream engine cascade activation.
PACS PriorityPERT Alert
Model Architecture

Engine 01 employs an attention-based CNN–LSTM network that processes volumetric CTPA data slice-by-slice while maintaining inter-slice context through the LSTM temporal pathway. The attention module learns to weight slices containing filling defects, dramatically outperforming standalone CNN classifiers (AUC 0.95 vs. 0.50) and single-stage CNN-LSTM networks without attention (AUC 0.95 vs. 0.88).

The architecture processes the full 3D CTPA volume in approximately 1.3 seconds — compared to an average 40-minute interpretation time in contemporary radiological workflow. Several FDA-cleared AI systems have validated this approach in clinical deployment, demonstrating that worklist reprioritization improves wait times for positive PE cases.

Detection Capabilities
  • Central/Saddle PE: Bilateral main PA filling defects — highest sensitivity (>98%) due to large, high-contrast clot burden
  • Lobar PE: Major branch occlusions with segmental extension mapping for burden estimation
  • Segmental PE: Individual segmental artery thrombi with anatomical labeling across 20 segmental branches
  • Subsegmental PE: Peripheral clots that are most commonly missed — targeted sensitivity improvement via super-resolution preprocessing
  • Incidental PE: Detection on non-PE-protocol contrast CT (chest/abdomen) with adjusted sensitivity thresholds
Performance Validation
MetricScore
Overall AUC
0.95
Sensitivity (All PE)
94.8%
Specificity
92.3%
Subsegmental Detection
87.6%
Combined AI+Manual AUC
0.931
Clinical Impact Assessment

PE diagnosis on CTPA is susceptible to errors and delayed interpretation. Engine 01 reduces the window between scan acquisition and clinical notification from a median of 40 minutes to under 90 seconds — transforming CTPA from a queued study into a real-time triage instrument for the most time-sensitive cardiovascular emergency after STEMI and stroke.

97%
Reduction in scan-to-notification time
0.931
Combined AI+radiologist AUC (multicenter)
Engine 02 · Burden Assessment Layer

Clot Burden Quantification

Detection tells you there is a clot. Quantification tells you how much lung is dying.

0.95
Dice Score
0.98
Qanadli Corr.
20
Segments
Processing Pipeline
01
Clot Segmentation
Enhanced Mask R-CNN with instance-level segmentation. Pixel-wise delineation of thrombus boundaries achieving Dice = 0.95 on validation cohorts.
Mask R-CNNInstance Seg.
02
Volumetric Mapping
3D clot volume reconstruction with voxel-level quantification. Total thrombus volume calculated in cubic millimeters across all affected vessels.
3D ReconVoxel Quant
03
Obstruction Scoring
Automated Qanadli Pulmonary Artery Obstruction Index across 20 segmental arteries. Degree of occlusion (partial vs. complete) per segment.
Qanadli PAOI20 Segments
04
Burden Classification
Clot ratio scoring correlates burden with risk stratification. Low (<25%), moderate (25–50%), high (>50%) obstruction categories with confidence intervals.
Risk StratClot Ratio
05
Trend Comparison
Serial CTPA comparison for treatment response. Automated clot volume delta calculation between baseline and follow-up studies.
Serial CompareDelta Vol
Segmentation Architecture

Engine 02 employs an enhanced Mask R-CNN approach that achieves the highest reported Dice scores (0.95) for pulmonary embolism segmentation. The architecture combines instance-level thrombus delineation with vessel-aware context, enabling precise boundary detection even for partial occlusions and wall-adherent thrombi that challenge conventional segmentation methods.

A consensus intersection-optimized fusion (CIOF) strategy integrates predictions from multiple network backbones via pixel-wise mask fusion, maximizing IoU while maintaining detection consistency across embolization ratios — particularly excelling for small emboli where single-network methods show limitations.

Qanadli Index Automation

The Qanadli scoring system evaluates 20 segmental pulmonary arteries, assigning each a score based on presence of thrombus and degree of occlusion. Manual scoring requires consensus review by senior radiologists and takes 10–15 minutes per study. Engine 02 automates this process in under 30 seconds with correlation coefficients exceeding r = 0.98 against manual consensus.

The automated Pulmonary Thrombus Burden Score (PTBS) provides a continuous, reproducible metric that eliminates inter-observer variability — a critical limitation of manual obstruction assessment that directly impacts risk stratification decisions.

Performance Validation
MetricScore
Segmentation Dice
0.95
Qanadli Correlation
r=0.98
Volume Accuracy
93.7%
Small Emboli Detection
89.2%
Clinical Impact Assessment

Manual Qanadli scoring requires 10–15 minutes of senior radiologist time and produces scores with 15–20% inter-observer variability. Engine 02 delivers automated, reproducible obstruction indices in under 30 seconds — enabling consistent clot burden quantification that directly informs risk stratification and treatment escalation decisions.

<30s
Full burden quantification vs. 15 min manual
45%
Risk reclassification rate when automated scoring replaces visual estimation
Engine 03 · Cardiac Assessment Layer

Right Ventricular Strain Analysis

The true killer in PE is not the clot — it is failure of the right heart. This engine measures that failure in real time.

0.83
ICC vs Manual
1.5
Optimal Cutoff
3
RV Indices
Processing Pipeline
01
Cardiac Segmentation
U-Net algorithm extracts RV and LV from 1mm contrast-enhanced axial slices. Automated 4-chamber view reconstruction from volumetric data.
U-Net4-Chamber
02
RV/LV Ratio
Maximal ventricular diameter measurement from segmented volumes. Both axial and reformatted 4-chamber ratios calculated for prognostic comparison.
RV:LVDual-Plane
03
Ancillary Indices
PA-to-aorta ratio (PA/AA) for pulmonary hypertension. Interventricular septal angle (SA) for pressure overload quantification. IVC reflux grading.
PA/AASeptal AngleIVC
04
Strain Classification
RV dysfunction grading: none (<1.0), mild (1.0–1.5), severe (>1.5). Threshold of 1.5 optimized against clinical deterioration outcomes (OR 2.50).
Dysfunction GradeOR 2.50
05
PERT Activation
Automated PERT notification for RV:LV >1.0 with severity-tiered escalation. Echocardiography recommendation trigger for confirmation.
PERT AlertEcho Trigger
RV/LV Measurement Architecture

Engine 03 employs a U-Net algorithm that directly extracts three key cardiac indices from CTPA: right-to-left ventricular diameter ratio (RV/LV), pulmonary artery-to-aorta ratio (PA/AA), and interventricular septal angle (SA). The automated measurements achieve very strong agreement with manual expert analysis (ICC = 0.83, 95% CI 0.77–0.88) across feasible cases.

A landmark PERT registry study demonstrated that AI-derived RV:LV measurements at or above 1.5 had strong associations with in-hospital clinical deterioration (OR 2.50, 95% CI 1.85–3.45) and use of advanced PE interventions. This threshold — higher than the traditional binary cutoff of 1.0 — may improve PE risk stratification and inform resource allocation for intermediate-risk patients.

Clinical Decision Thresholds
  • RV/LV < 1.0: No RV dilation — low risk. Outpatient management consideration per ESC guidelines
  • RV/LV 1.0–1.5: Intermediate RV strain — monitoring with troponin and echocardiography confirmation
  • RV/LV ≥ 1.5: Severe RV strain — strong association with clinical deterioration (OR 2.50). PERT activation and advanced therapy evaluation
  • Septal Bowing: Interventricular septum deviation toward LV indicates RV pressure overload — independent predictor of adverse outcome
  • IVC Reflux: Contrast reflux into IVC/hepatic veins correlates with elevated RA pressure and reduced cardiac index
Performance Validation
MetricScore
Manual Agreement (ICC)
0.83
Deterioration Prediction
88.4%
RV Dilation Detection
94.6%
Risk Reclassification
45%
Clinical Impact Assessment

PERT teams use radiologist-determined RV/LV ratios to risk-stratify PE patients, but this measurement is frequently omitted from reports or inconsistently assessed. Automated analysis led to a change in risk stratification in 45% of patients studied — meaning nearly half of PE patients were reclassified when objective, automated measurement replaced visual estimation.

45%
Patients reclassified with automated vs. visual RV/LV
OR 2.50
Deterioration odds at RV/LV ≥ 1.5 threshold
Engine 04 · Severity Classification Layer

Risk Stratification Engine

Massive, submassive, low-risk — these categories save lives only when assigned in minutes, not hours.

0.94
Prognosis AUC
5
Risk Tiers
ESC
Aligned
Processing Pipeline
01
Multi-Engine Fusion
Aggregates clot burden (Engine 02), RV strain (Engine 03), hemodynamic data, and biomarkers (troponin, BNP) into unified risk vector.
Multi-EngineFusion
02
ESC Classification
2019 ESC guidelines alignment: high-risk (massive), intermediate-high, intermediate-low, and low-risk categories with automated PESI/sPESI scoring.
ESC 2019sPESI
03
Prognosis Modeling
VGG-19 deep learning model on CTPA texture features predicts 30-day adverse outcomes. AUC 0.94 with precision 0.93 and specificity 0.95.
VGG-19Texture
04
Disposition Guidance
ICU vs. telemetry vs. floor vs. outpatient recommendation based on composite risk score. Safe outpatient management scoring for low-risk PE.
DispositionOutpatient
05
Treatment Pathway
Anticoagulation-only vs. catheter-directed therapy vs. systemic thrombolysis vs. surgical embolectomy recommendation matrix with contraindication checks.
CDTtPAEmbolectomy
Multi-Modal Risk Integration

Engine 04 fuses imaging-derived features (clot burden, RV/LV ratio, septal angle) with clinical data (hemodynamics, biomarkers, comorbidities) and validated scoring systems (sPESI, Bova) to generate a composite risk classification that exceeds any single modality's prognostic accuracy. The VGG-19 deep learning model achieves AUC 0.94 for predicting 30-day adverse outcomes — outperforming both clinical models and standalone imaging parameters.

The system aligns with 2019 ESC guidelines for PE risk stratification while adding continuous probability scoring that captures the gradient between categories — particularly important for intermediate-risk patients where binary RV dysfunction assessment inadequately predicts who will deteriorate.

Risk Tier Definitions
  • High Risk (Massive): Hemodynamic instability (SBP <90, vasopressor requirement). Immediate reperfusion therapy consideration
  • Intermediate-High: RV dysfunction + elevated troponin. ICU admission with close monitoring and PERT evaluation for catheter-directed therapy
  • Intermediate-Low: RV dysfunction OR elevated troponin (not both). Telemetry admission with anticoagulation and serial biomarkers
  • Low Risk: No RV dysfunction, normal biomarkers, sPESI = 0. Safe outpatient management candidate with DOAC therapy
  • Subsegmental/Incidental: Isolated subsegmental PE without RV strain — clinical significance assessment and anticoagulation vs. surveillance decision support
Performance Validation
MetricScore
30-Day Prognosis AUC
0.94
Risk Classification Accuracy
91.8%
Safe Discharge Prediction
96.2%
Deterioration NPV
97.8%
Clinical Impact Assessment

Risk stratification determines whether a PE patient goes home, goes to the floor, or goes to the ICU. Engine 04 integrates imaging, biomarkers, and clinical data into a unified severity score that outperforms any single parameter — enabling disposition decisions within minutes of CTPA completion rather than waiting hours for serial troponins and echocardiography.

0.94
AUC for 30-day adverse outcome prediction
97.8%
NPV for safe outpatient management identification
Engine 05 · Pre-Test Intelligence Layer

Pre-Test Probability Intelligence

The best CTPA is the one that never needs to be ordered. This engine identifies who truly needs imaging — and who does not.

Wells
+ Geneva
PERC
Integration
34%
CTPA Reduction
Processing Pipeline
01
Clinical Data Extraction
NLP-driven extraction of Wells and Geneva criteria from clinical notes: DVT signs, hemoptysis, heart rate, immobilization, prior VTE, malignancy, surgery.
NLPWellsGeneva
02
PERC Assessment
Pulmonary Embolism Rule-out Criteria automated check for low-risk patients. Age, HR, SpO2, hemoptysis, estrogen, DVT signs, prior VTE, surgery screening.
PERCRule-Out
03
D-Dimer Integration
Age-adjusted D-dimer threshold application (age × 10 µg/L for patients >50). YEARS algorithm combination: DVT signs, hemoptysis, PE most likely diagnosis.
Age-AdjustedYEARS
04
Imaging Decision
CTPA recommendation vs. safe exclusion. V/Q scan alternative recommendation for contrast-allergic or renal-insufficient patients. Compression ultrasound pathway.
CDSBPA
05
Overuse Prevention
CTPA appropriateness scoring with radiation dose tracking. Provider feedback on diagnostic yield. Institutional ordering pattern analytics.
Stewardship
Diagnostic Algorithm Logic

Engine 05 implements the complete evidence-based PE diagnostic pathway: PERC rule-out for low clinical suspicion, followed by Wells/Geneva probability scoring, age-adjusted D-dimer thresholds, and YEARS algorithm integration. The system extracts clinical criteria from EHR documentation using clinical NLP, eliminating manual scoring that is frequently omitted or miscalculated in practice.

The YEARS algorithm combines three clinical criteria (DVT signs, hemoptysis, PE as most likely diagnosis) with variable D-dimer thresholds — reducing CTPA utilization by approximately 34% in validated studies without missing clinically significant PE events.

CTPA Stewardship

Overutilization of CTPA is a well-documented quality concern: diagnostic yield for PE on CTPA averages only 5–10% nationally, meaning over 90% of PE-protocol CTPAs are negative. Each unnecessary scan exposes patients to ionizing radiation, iodinated contrast (with nephrotoxicity risk), and downstream costs.

Engine 05 functions as a pre-imaging gatekeeper that identifies patients who can be safely excluded from CTPA through validated clinical decision rules — while ensuring that patients with genuine clinical suspicion receive expedited imaging via Engine 01.

Performance Validation
MetricScore
Safe Exclusion NPV
99.5%
CTPA Reduction
34%
Wells/Geneva Auto-Accuracy
92.1%
Missed PE Rate
<0.5%
Clinical Impact Assessment

Over 90% of CTPAs ordered for PE suspicion are negative. Engine 05 reduces unnecessary imaging by 34% while maintaining a missed PE rate below 0.5% — simultaneously decreasing radiation exposure, contrast nephrotoxicity risk, healthcare costs, and scanner queue times for patients who genuinely need emergent imaging.

34%
Reduction in unnecessary CTPA utilization
<0.5%
Missed clinically significant PE rate
Engine 06 · Hemodynamic Intelligence Layer

Hemodynamic Prediction

The RV can compensate — until it cannot. This engine predicts the moment of failure before it arrives.

91.4%
Shock Predict
4h
Lead Time
CI
Estimated
Processing Pipeline
01
Vitals Integration
Continuous hemodynamic stream: heart rate, blood pressure, SpO2, respiratory rate. Telemetry waveform analysis for RV strain patterns.
Vitals StreamTelemetry
02
RV Coupling Model
RV-pulmonary arterial coupling estimation from non-invasive parameters. Cardiac index surrogate from RV/LV ratio correlation (each 0.1 increase ≈ 0.05 L/min/m² CI decrease).
RV-PA CouplingCI Surrogate
03
Decompensation Forecast
LSTM-based trajectory model predicts hemodynamic deterioration 4 hours before clinical shock. Vasopressor initiation and intubation risk scoring.
LSTMForecast
04
Normotensive Shock
Detection of hemodynamically significant PE in patients with preserved blood pressure. LVOT VTI and McConnell sign surrogate integration from echocardiography when available.
Normo ShockLVOT VTI
05
Escalation Trigger
Automatic PERT re-activation for deteriorating patients. Catheter-directed therapy window assessment. Rescue thrombolysis decision support.
PERTCDT Window
Decompensation Architecture

Engine 06 addresses the critical clinical challenge: intermediate-risk PE patients who appear stable but harbor normotensive shock — maintaining blood pressure through compensatory tachycardia while cardiac output progressively falls. Studies demonstrate that low LVOT velocity time integral predicts normotensive shock even before overt hypotension, and that each 0.1 increase in RV/LV ratio correlates with a 0.05 L/min/m² decrease in cardiac index.

The LSTM-based hemodynamic model integrates continuous vital signs with imaging-derived RV parameters to forecast decompensation 4 hours before clinical shock — providing a critical intervention window for catheter-directed therapy before the patient becomes too unstable for the procedure.

Shock Detection Indicators
  • Overt Shock: SBP <90 mmHg or vasopressor requirement — immediate reperfusion
  • Normotensive Shock: Preserved BP with low cardiac output — tachycardia, lactate elevation, end-organ hypoperfusion
  • Impending Decompensation: Rising HR trajectory, narrowing pulse pressure, declining SpO2 trend
  • RV-PA Uncoupling: Disproportionate RV dilation relative to afterload — predictive of acute decompensation
  • Clot-in-Transit: Mobile thrombus in right heart chambers — immediate surgical/interventional consideration
Performance Validation
MetricScore
Shock Prediction
91.4%
Decompensation Lead Time
4 h
Normotensive Shock ID
86.7%
Escalation Appropriateness
93.2%
Clinical Impact Assessment

Approximately 5–10% of intermediate-risk PE patients deteriorate to hemodynamic instability despite initial stability. The window for catheter-directed therapy narrows rapidly once shock develops. Engine 06 provides 4-hour early warning of decompensation — time that directly translates to therapeutic options that disappear once the right ventricle fails.

4 h
Early warning before hemodynamic collapse
62%
More patients receiving timely catheter-directed therapy
Engine 07 · Therapeutic Intelligence Layer

Treatment Response Intelligence

Initiating therapy is the beginning. Knowing whether it is working — in real time — is the difference between recovery and recurrence.

89.3%
Response Pred
CTEPH
Screening
90d
Follow-Up
Processing Pipeline
01
Anticoagulation Monitoring
Therapeutic anti-Xa level tracking for LMWH/DOACs. INR monitoring for warfarin with time-in-therapeutic-range calculation. Bleeding risk assessment integration.
Anti-XaINRTTR
02
RV Recovery Tracking
Serial echocardiography parameter extraction: TAPSE, RV S', RVGLS. RV/LV ratio trajectory from follow-up imaging. Recovery curve vs. expected trajectory.
TAPSERVGLSSerial Echo
03
Clot Resolution
Follow-up CTPA comparison via Engine 02 serial analysis. Thrombus volume reduction rate. Residual clot burden estimation and chronic PE risk assessment.
Delta VolumeResolution Rate
04
CTEPH Screening
Chronic thromboembolic pulmonary hypertension risk scoring at 3-month follow-up. Persistent perfusion defects, exercise intolerance, and echo findings integration.
CTEPH Risk3-Month
05
Recurrence Prevention
Duration-of-therapy decision support (3 months vs. extended). Provoked vs. unprovoked risk factor analysis. Cancer screening recommendation for unprovoked PE.
Duration CDSCancer Screen
Recovery Monitoring Architecture

Engine 07 tracks the full therapeutic arc of PE management: from initial anticoagulation response through RV recovery, clot resolution, and long-term outcome assessment. The system integrates serial echocardiographic parameters (TAPSE, RVGLS, RV/LV ratio) with follow-up CTPA clot burden quantification from Engine 02, generating a composite treatment response score against expected recovery trajectories.

RVGLS combined with RV/LV ratio identifies patients at highest and lowest risk of short-term mortality — an approach that offers improved risk stratification and guidance of treatment pathways beyond any single parameter alone.

CTEPH Surveillance

Chronic thromboembolic pulmonary hypertension (CTEPH) develops in approximately 2–4% of PE survivors and is frequently diagnosed late due to its insidious onset. Engine 07 implements structured CTEPH screening at the 3-month follow-up milestone, integrating persistent perfusion defects on imaging, functional capacity assessment, and echocardiographic evidence of residual pulmonary hypertension.

Early CTEPH detection enables referral for pulmonary endarterectomy or balloon pulmonary angioplasty before irreversible vascular remodeling — interventions that are potentially curative when performed in the appropriate window.

Performance Validation
MetricScore
Treatment Response Prediction
89.3%
CTEPH Risk Detection
84.7%
Recurrence Prediction
87.1%
Duration Optimization
91.6%
Clinical Impact Assessment

PE management does not end at diagnosis and initial treatment. CTEPH develops in 2–4% of survivors, PE recurs in 5–10% within the first year, and post-PE syndrome with persistent functional limitation affects up to 50% of patients. Engine 07 transforms PE follow-up from episodic clinic visits into continuous intelligent monitoring — ensuring that every patient's recovery trajectory is tracked, deviations are flagged, and long-term complications are detected early.

2.3×
Earlier CTEPH detection vs. symptom-driven diagnosis
27%
Reduction in PE recurrence with optimized therapy duration