Skip to content

Latest commit

 

History

1 Commit

Folders and files

Repository files navigation

MicroVasc: Computational Tumor Angiogenesis & Vascular Normalization Simulator

CI Python Version License: MIT Scientific Computing

MicroVasc is a multiscale computational oncology library for modeling tumor microvascular sprouting, endothelial Delta-Notch lateral inhibition, network hemodynamics, and clinical anti-VEGF vascular normalization therapy.


1. Mathematical & Biophysical Architecture

flowchart TD
    subgraph Microenvironment["Biochemical Transport (Reaction-Diffusion)"]
        O2["Oxygen Field pO2(x,y)"] -->|Hypoxia Induction| HIF["HIF-1alpha Activation"]
        HIF -->|Secretion| VEGF["VEGF Chemokine Gradient grad(C_V)"]
        VEGF -->|Receptor Binding| VEGFR2["Endothelial VEGFR2"]
    end

    subgraph Cellular["Cellular Signaling & Sprouting"]
        VEGFR2 -->|Dll4 Synthesis| Delta["Delta (Dll4) Expression"]
        Delta -->|Lateral Inhibition| Notch["Notch1 Activation in Neighbors"]
        Notch -->|Downregulation| Phenotype{"Phenotype: Tip vs Stalk"}
        Phenotype -->|High Delta| TipCell["Chemotactic Tip Cell Migration"]
        Phenotype -->|High Notch| StalkCell["Proliferating Stalk Lumen"]
        TipCell -->|Snail Trail| Branching["Sprouting & Anastomosis Loops"]
    end

    subgraph Hemodynamics["Microvascular Hemodynamics"]
        Branching --> Circuit["Closed Capillary Graph Network"]
        Circuit --> Kirchhoff["Kirchhoff Nodal Pressure Solver"]
        Kirchhoff --> Poiseuille["Poiseuille Flow & Shear Stress tau_w"]
        Poiseuille --> Perfusion["Functional Blood Perfusion & O2 Delivery"]
        Perfusion -.->|Negative Feedback| O2
    end

    subgraph Clinical["Clinical Oncology & Therapy"]
        AntiVEGF["Anti-VEGF Therapy (e.g. Bevacizumab)"] -->|Prune Immature Non-perfused Sprouts| Normalization["Vascular Normalization Window"]
        Normalization --> ReducedIFP["Decreased Interstitial Fluid Pressure"]
        Normalization --> ImprovedChemo["Optimized Chemotherapy & Radiation Timing"]
    end
Loading

1.1 Coupled Reaction-Diffusion Transport Equations

The spatial transport of vascular endothelial growth factor ($C_V$) and dissolved tissue oxygen ($C_{O2}$) is governed by the two-dimensional reaction-diffusion partial differential equations:

$$\frac{\partial C_V}{\partial t} = D_V \nabla^2 C_V - \gamma_V C_V + S_V(x, y) \cdot \mathcal{H}(C_{\text{hypoxic}} - C_{O2}) - k_{\text{bind}} B_{\text{anti-VEGF}} C_V$$

$$\frac{\partial C_{O2}}{\partial t} = D_{O2} \nabla^2 C_{O2} - \gamma_{O2} C_{O2} + S_{\text{vessel}}(x, y) - \alpha_{\text{cell}} C_{O2}$$

Where:

  • $D_V, D_{O2}$ are the effective tissue diffusion coefficients.
  • $\mathcal{H}$ is the Heaviside step function inducing VEGF synthesis upon cellular hypoxia ($pO_2 < 10\text{ mmHg}$).
  • $k_{\text{bind}} B_{\text{anti-VEGF}} C_V$ models pharmacological neutralization by monoclonal antibodies (e.g., Bevacizumab).

1.2 Delta-Notch Lateral Inhibition Dynamics

Endothelial cells along the sprouting front differentiate into migratory tip cells or proliferative stalk cells via Notch1-Delta-like 4 (Dll4) juxtacrine lateral inhibition:

$$\frac{d D_i}{dt} = v_D \frac{1}{1 + (N_i / k_N)^m} \cdot \frac{R_i^p}{k_R^p + R_i^p} - \mu_D D_i$$

$$\frac{d N_i}{dt} = v_N \frac{\langle D_{\text{neighbors}} \rangle^k}{k_D^k + \langle D_{\text{neighbors}} \rangle^k} - \mu_N N_i$$

Cells expressing $D_i \ge \theta_{\text{tip}}$ acquire the tip cell phenotype, extending filopodia along $\nabla C_V$.

1.3 Microvascular Hemodynamics & Poiseuille Flow

In newly anastomosed capillary loops, blood flow rate $Q_{ij}$ between junctions $i$ and $j$ obeys Poiseuille's law:

$$Q_{ij} = G_{ij} (P_i - P_j), \quad G_{ij} = \frac{\pi R_{ij}^4}{8 \mu L_{ij}}$$

Enforcing Kirchhoff's current conservation at all internal vascular junctions yields:

$$\sum_{j \in \mathcal{N}(i)} G_{ij} (P_i - P_j) = 0 \implies \mathbf{L} \mathbf{P} = \mathbf{b}$$

Endothelial wall shear stress is computed as:

$$\tau_w = \frac{4 \mu Q}{\pi R^3}$$

1.4 Clinical Vascular Normalization Index

Following Rakesh K. Jain's vascular normalization hypothesis, anti-angiogenic therapy selectively prunes hyperpermeable, non-perfused immature sprouts, reducing Interstitial Fluid Pressure (IFP) and decreasing the tumor hypoxic fraction ($HF_{10}$):

$$N_{\text{score}} = 0.35 \cdot \left(\frac{N_{\text{perfused}}}{N_{\text{total}}}\right) + 0.35 \cdot (1 - HF_{10}) + 0.15 \cdot \left(\frac{1}{T}\right) + 0.15 \cdot \left(1 - \frac{\text{IFP}}{25}\right)$$

When $N_{\text{score}} \ge 0.65$ and $HF_{10} < 0.30$, the tumor is within the vascular normalization window, providing optimal microvascular perfusion for cytotoxic chemotherapy and radiation delivery.


2. Installation

Requires Python 3.11, 3.12, or 3.14.

git clone https://github.com/ZengLLQ/micro-vasc.git
cd micro-vasc
pip install -e ".[dev]"

3. Command-Line Interface (CLI)

3.1 Run Angiogenic Sprouting Simulation

micro-vasc simulate --grid 48 --steps 35 --radius 10

3.2 Test Delta-Notch Lateral Inhibition Patterning

micro-vasc delta-notch --cells 12 --vegf 1.5 --steps 100

3.3 Audit Anti-VEGF Vascular Normalization Therapy

micro-vasc normalize --grid 48 --dose 25.0

4. Python API Usage

from micro_vasc import AngiogenesisSimulator
from micro_vasc.visualizer import AsciiVisualizer

# 1. Initialize coupled multiscale simulator
sim = AngiogenesisSimulator(grid_size=48, tumor_radius=10, rng_seed=42)

# 2. Advance 30 temporal steps of sprouting & anastomosis
sim.run(steps=30)

# 3. Render real-time spatial tissue map
print(AsciiVisualizer.render_spatial_map(sim.env, sim.network, display_size=24))

# 4. Evaluate clinical normalization audit
metrics = sim.audit_normalization()
print(AsciiVisualizer.render_clinical_report(metrics))

# 5. Administer targeted anti-VEGF therapy (e.g. 20 nM Bevacizumab)
pruned_vessels = sim.therapy.prune_immature_vessels(anti_vegf_dose_nm=20.0)
print(f"Pruned {pruned_vessels} immature sprouts.")

5. Verification & Testing

Run the full pytest suite locally:

pytest -v

All unit and integration tests run across Ubuntu, macOS, and Windows runners in GitHub Actions CI.


6. License

This project is licensed under the MIT License - see the LICENSE file for details.