ConductScreen
External Validation

Ethanol-Exposed Organoid Reanalysis

ConductScreen pipeline detects substance-induced morphological changes in published data without parameter adjustment

Study Source

Adams JW et al. Impact of alcohol exposure on neural development and network formation in human cortical organoids. Mol Psychiatry 28, 1571-1584 (2023). DOI

Exposure: ~20 mM steady-state ethanol
Model: Human cortical organoids (iPSC-derived)
Sample size: N=4 control, N=5 ethanol

Why This Matters for ConductScreen

Our Phase 1 data demonstrates that ConductScreen discriminates genetic disease models from wildtype (27/48 significant comparisons). But the target context of use is detecting substance-induced changes. This reanalysis bridges that gap: applying the same pipeline to published ethanol exposure data shows it detects chemical perturbation — even from figure-extracted images at small N.

Phase 1: Genetic disease detection (demonstrated)
Bridging evidence: Chemical sensitivity (this page)
Phase 2: Full dose-response validation

Automated Segmentation of External Images

The ConductScreen pipeline processed Adams et al. Figure 1b images without parameter adjustment. Green contours show automated segmentation boundaries.

Control (N=4)

4.7k
C1
4.8k
C2
2.6k
C3
4.9k
C4

Ethanol (N=5)

4.0k
E1
2.3k
E2
4.7k
E3
2.5k
E4
1.6k
E5

Circle size proportional to organoid area (px²). Control mean: 4,237 px² • Ethanol mean: 3,011 px² (↓29%). Green contour overlays available on live server.

Key Findings

Ethanol-exposed organoids show consistent size reduction across multiple morphological features.

Area

-29%

p = 0.085

Trend toward smaller organoids (directional evidence)

Perimeter

-21%

p = 0.049*

Statistically significant reduction in boundary length

Equiv. Diameter

-17%

p = 0.085

Shape-independent size reduction confirms growth arrest

Circularity

+11%

p = 0.219

Trend toward rounder shapes, potentially reflecting reduced complexity

Feature-by-Feature Comparison

Mann-Whitney U tests (two-sided) comparing control vs. ethanol-exposed organoids. All spatial measurements in pixel units (calibration unknown).

FeatureControl (N=4)Ethanol (N=5)Changep-value
Area(px²)4237.1 ± 962.73010.6 ± 1128.4-28.9%0.085
Perimeter(px)258.4 ± 11.6204.8 ± 38.2-20.8%0.049 *
Circularity0.8 ± 0.10.9 ± 0.0+10.7%0.219
Equiv. Diameter(px)72.9 ± 9.060.8 ± 11.6-16.6%0.085
Major Axis(px)74.4 ± 7.161.2 ± 11.5-17.7%0.085
Solidity1.0 ± 0.01.0 ± 0.0+2.6%0.623

* Statistically significant at p < 0.05. Mann-Whitney U test (two-sided, uncorrected). Small N — interpret as directional evidence.

Consistency with Published Findings

Adams et al. Report

“Ethanol exposure led to growth arrest, reduced neuronal differentiation, and disrupted network formation in cortical organoids.”

ConductScreen Detects

Area ↓29%, Perimeter ↓21% (p=0.049), Equivalent Diameter ↓17% — quantitative confirmation of growth arrest using automated morphological analysis.

Pipeline generalization: No parameters were adjusted between the primary analysis (Schröter et al. iOrganBio data) and this external validation (Adams et al. data). The same segmentation and feature extraction pipeline processes both datasets identically.

Limitations & Caveats

  • Images extracted from published figure, not original raw microscopy
  • Pixel calibration unknown — all spatial measurements in pixel units, not μm
  • Small sample size (N=4 control, N=5 ethanol) — interpret as directional evidence
  • Single ethanol concentration (~20 mM steady-state)
  • No dose-response relationship assessable from this dataset

What Phase 2 Adds

Phase 2 replaces this proof-of-concept reanalysis with a purpose-built validation study:

5 ethanol concentrations (0, 10, 25, 50, 100 mM) for dose-response curves
Valproic acid (VPA) as second reference substance with known DNT
N ≥ 30 organoids per condition for statistical power
Pre-registered statistical analysis plan (SAP)
Raw microscopy at known magnification for calibrated measurements
Immunofluorescence validation of morphological endpoints