---
title: 'MCMeshGAN: 3D Aneurysm Growth Predictor'
url: https://www.emergentmind.com/topics/mcmeshgan
type: topic
---

# MCMeshGAN: 3D Aneurysm Growth Predictor

MCMeshGAN, short for **Multimodal Conditional MeshGAN**, is a **multimodal conditional mesh-to-mesh generative adversarial network** proposed for **personalized 3D prediction of thoracic aortic aneurysm growth**. It is presented as the first method of this kind for aneurysm progression modeling directly on 3D surface meshes, rather than on 2D images or handcrafted geometric summaries. The model takes a source aneurysm mesh together with **age**, **sex**, and a chosen **target time interval**, and generates a patient-specific target mesh at that time, supporting both retrospective reconstruction and prospective prediction [2508.19862].

## 1. Clinical problem and modeling objective

MCMeshGAN is designed for the problem of **predicting how an individual patient’s aortic aneurysm will change over time**, with the stated goal of improving surveillance and intervention timing. In the formulation described for thoracic aortic aneurysm progression, current practice often relies on **manual diameter measurements** on follow-up imaging and threshold-based clinical rules. The method is motivated by the observation that aneurysm progression is fundamentally a **3D anatomical process**, whereas diameter-based assessment reduces it to a sparse, mostly 2D surrogate [2508.19862].

The paper defines the task as **mesh-to-mesh 3D anatomical progression**: given a baseline aortic surface mesh, the model generates a future mesh for the same patient. The stated technical difficulty is that aneurysm growth includes both **subtle local deformations** and **broader global anatomical changes** on **non-Euclidean** meshes. Standard CNNs are therefore not naturally applicable, while graph-based methods, although topology-aware, are described as vulnerable to **over-smoothing** in deep GCNs, which blurs local distinctions and harms fine geometric fidelity [2508.19862].

Within this setting, MCMeshGAN fills a specific gap: prior work, as characterized in the paper, had not provided a **conditional, multimodal, mesh-to-mesh generative model** for personalized aneurysm growth prediction that can simultaneously use **3D geometry**, **clinical attributes**, and a **target future time interval**, while preserving both local and global anatomy. This suggests that the method should be understood not merely as a mesh generator, but as a trajectory model for patient-specific anatomical evolution [2508.19862].

## 2. Conditional formulation and geometric representation

The model’s input-output relation is stated explicitly. During training, given a source mesh \(m_s\) with source clinical data \((a_s, g_s, t_s)\), the model predicts a target mesh \(m_t\) from another time point \(t_t\), using

$$
time_{interval} = t_t - t_s.
$$

At inference, given a source aneurysm mesh \(m_s\), age \(a_s\), sex \(g_s\), and a chosen target interval, the generator predicts

$$
m_t = G(m_s, a_s, g_s \mid time_{interval}).
$$

Positive intervals indicate future prediction; negative intervals indicate past reconstruction [2508.19862].

In this formulation, **multimodal** denotes the combination of **3D mesh geometry** with **clinical variables**, specifically **age**, **sex**, and the **target time interval**. **Conditional** denotes explicit control by these auxiliary variables, especially the requested temporal offset. **Mesh-to-mesh GAN** denotes a generative adversarial translation framework in which a source mesh is transformed into a predicted target mesh, while adversarial training is used to encourage realism [2508.19862].

The anatomical representation is a **triangular 3D surface mesh** extracted from CT. The preprocessing pipeline comprises **CT segmentation** using the pre-trained **TotalSegmentator** model, **manual correction** of segmentation masks, **mesh extraction** with the **marching cubes** algorithm in the IRTK library, **rigid alignment**, **non-rigid surface registration**, and **standardization to 10,000 vertices** for every mesh. These operations enforce **one-to-one vertex correspondence** and a common topology across samples. The paper does not describe prediction of a displacement field; it describes prediction of the target mesh itself, which under this representation is effectively prediction of **future mesh coordinates** [2508.19862].

A strong implication of this design is that the learning problem is simplified from arbitrary topology generation to anatomical progression under fixed mesh structure. This also explains why the paper reports that \(L_1\) loss works better than Chamfer Distance under the corresponding-vertex regime [2508.19862].

## 3. Architecture: condition branch, KCN branch, and GCN branch

MCMeshGAN is organized around three main branches: a **condition branch**, a **local KNN-based convolutional network (KCN)** branch, and a **global graph convolutional network (GCN)** branch, followed by a fusion module that integrates features from these branches [2508.19862].

The **condition branch** encodes **age**, **sex**, and **target time interval**. The paper introduces a **Text2Vector** encoder based on **ordinal binary encoding**, where each clinical attribute is represented by a \(100 \times 1\) binary vector. The most explicit definition is given for temporal conditioning:

$$
time\_interval\_vector \in \mathbb{R}^{1\times 100},
$$

with

$$
time\_interval\_vector[1:50+time\ interval]=0,
$$

$$
time\_interval\_vector[50+time\ interval+1:100]=1.
$$

Index 50 serves as the dividing point between past and future. Age and sex are also transformed to vectors by the same Text2Vector ordinal binary encoding scheme, although no explicit formulas are given for those variables. This branch provides both personalization and temporal control [2508.19862].

The **local KCN branch** is introduced to preserve fine local detail that deep GCNs may smooth away. Starting from mesh vertices

$$
V \in \mathbb{R}^{N \times d},
$$

a CNN-based feature extractor maps raw vertex coordinates into per-vertex features:

$$
\mathcal{F}(V) = \mathcal{F}_{CNN}(V), \qquad \mathcal{F}(V)\in \mathbb{R}^{N\times d_f}.
$$

For each query vertex \(q\), the branch computes its \(K\)-nearest neighbors,

$$
\mathcal{N}(q) = \{v_{i_1}, v_{i_2}, \ldots, v_{i_k}\},
$$

gathers the corresponding features,

$$
\mathcal{F}(\mathcal{N}(q)) = \{\mathcal{F}(v_{i_1}), \mathcal{F}(v_{i_2}), \ldots, \mathcal{F}(v_{i_k})\},
$$

and applies two parallel convolution modules, \(\mathcal{F}_{Center}\) and \(\mathcal{F}_{Nei}\), before vertex interaction and aggregation:

$$
\mathcal{F}_{local} = \mathcal F_M \Big( \mathcal F_C [ \mathcal{F}_{Nei}(\mathcal F (\mathcal{N}(q))), \mathcal{F}_{Center}(F(q)) ] \Big).
$$

The paper sets \(K=8\) empirically. This branch is explicitly linked to **fine-grained local deformations** and to accurate measurement of clinically important diameter change [2508.19862].

The **global GCN branch** treats the mesh as an undirected graph

$$
\mathcal G = (V, \xi),
$$

with connectivity encoded by a binary adjacency matrix

$$
A \in \mathbb{R}^{N \times N},
$$

where \(A_{ij}=1\) if vertices \(v_i\) and \(v_j\) are connected. The graph convolution is the standard Kipf–Welling layer:

$$
V' = \bar{D}^{-1/2}\bar{A}\bar{D}^{-1/2}V\Theta,
$$

with

$$
\bar{A} = A + I_N,
$$

and

$$
\bar{D}_{ii} = \sum_j \bar{A}_{ij}.
$$

The branch consists of **four graph convolutional blocks**, which expand the receptive field and encode the **global anatomical structure** of the aorta [2508.19862].

The fusion module is described conceptually rather than algebraically. The paper does not provide an explicit fusion equation or exact fusion operator, but it states that the final prediction is conditioned on the combined information from the KCN branch, the GCN branch, and the condition branch. The stated answer to GCN over-smoothing is architectural: a **dual-branch decomposition** in which GCN is used for global context and KCN for local detail, rather than a modified graph operator or explicit anti-smoothing penalty [2508.19862].

## 4. Training protocol, losses, and TAAMesh

The paper introduces **TAAMesh**, a longitudinal thoracic aortic aneurysm mesh dataset consisting of **590 CT scans / records** from **208 patients**, with **2 to 8 longitudinal scans** per patient. The modalities include **CT scans**, **3D meshes**, and **clinical data**. The reported clinical variables include **age**, **sex**, **maximum inscribed sphere (MIS) diameter**, and corresponding clinical reports. The data come from **Hammersmith Hospital, Imperial College London**, under Health Research Authority approval **23/HRA/3733**. The split is at the **patient level** with a **7:1:2 train/validation/test ratio** [2508.19862].

Training pairs are formed longitudinally from the **same patient**, using source mesh \(m_s\) at time \(t_s\) and target mesh \(m_t\) at time \(t_t\), with the interval \(time_{interval}=t_t-t_s\). Because the interval may be positive or negative, the same framework supports **prospective prediction** to future timepoints and **retrospective reconstruction** to earlier timepoints [2508.19862].

The implementation details reported in the paper are specific. The framework uses **PyTorch Geometric** for GCN baselines. Training is performed on an **Intel Xeon Silver 4114 CPU** and a single **Tesla V100 32GB** GPU, with **Adam**, learning rate

$$
\alpha = 0.0002,
$$

Adam parameter

$$
\beta_1 = 0.5,
$$

**1000** epochs, **batch size 1**, and \(K=8\) in the KNN module. The paper does not mention data augmentation [2508.19862].

The loss design is described only partially in symbolic terms. The paper states that three losses are evaluated: **\(L_1\) loss**, **Chamfer Distance (CD) loss**, and **adversarial loss**. It does **not** provide explicit formulas for these losses, nor the final weighted objective, nor weighting coefficients between losses. What is stated is that adversarial loss improves realism and anatomical plausibility, \(L_1\) works better than CD because meshes are non-rigidly registered into **one-to-one vertex correspondence**, and the best final model uses **\(L_1 +\) adversarial loss**. The paper also does **not** define additional regularizers such as Laplacian smoothness, edge-length preservation, normal consistency, explicit diameter loss, or deformation regularization [2508.19862].

The adversarial framework itself is only partially specified. MCMeshGAN is explicitly described as a GAN, and adversarial loss is reported to improve realism, but no detailed discriminator architecture is given in the text provided. The absence of a layer-by-layer discriminator description distinguishes MCMeshGAN from earlier mesh GANs such as MeshGAN, which used a BEGAN-style autoencoder discriminator built with Chebyshev spectral graph convolutions for fixed-topology 3D face generation [1903.10384].

## 5. Evaluation, ablations, and empirical findings

Evaluation uses geometric mesh metrics in millimeters—**MAE**, **CD**, and **HD**—together with a clinically meaningful metric, **MIS diameter error**, defined as the error in maximum inscribed sphere diameter estimation. The paper does not provide formal formulas for these metrics, but it states their respective roles: average pointwise or vertexwise error, set-to-set nearest-neighbor surface discrepancy, worst-case surface discrepancy, and clinical accuracy of maximum diameter estimation [2508.19862].

The compared baselines are **PointNet++**, **GCNConv**, **GCN2Conv**, **SSGConv**, and **PMLP**, each adapted with the same condition branch for fairness. The paper also evaluates variants of several baselines with **+Condition+KCN** to test whether the KCN module improves them [2508.19862].

On the test set, MCMeshGAN reports the best performance among all compared methods:

| Model | MAE / CD / HD | MIS Error |
|---|---|---|
| MCMeshGAN | \(1.285 \pm 0.993\) mm / \(1.831 \pm 1.251\) mm / \(9.815 \pm 5.131\) mm | \(2.887\) mm |
| SSGConv + Condition | \(1.875 \pm 3.371\) mm / \(2.495 \pm 2.280\) mm / \(13.091 \pm 9.107\) mm | \(3.916\) mm |
| GCN2Conv + Condition + KCN | \(1.605 \pm 3.516\) mm / \(2.228 \pm 2.457\) mm / \(12.161 \pm 9.534\) mm | \(3.385\) mm |

The paper emphasizes that MCMeshGAN best matches ground-truth geometry across the core metrics and achieves the **lowest MIS diameter error**, which it treats as particularly clinically meaningful. Compared with the strong KCN-augmented baseline **GCN2Conv + Condition + KCN**, the reported improvements are \(1.605 \rightarrow 1.285\) in MAE, \(2.228 \rightarrow 1.831\) in CD, \(12.161 \rightarrow 9.815\) in HD, and \(3.385 \rightarrow 2.887\) in MIS Error [2508.19862].

The ablation study attributes this result to several interacting factors. Full MCMeshGAN with age, sex, KCN, GCN, \(L_1\), and adversarial loss gives **MAE \(1.285 \pm 0.993\)**, **CD \(1.831 \pm 1.251\)**, and **HD \(9.815 \pm 5.131\)**. Removing age degrades performance to **MAE \(1.533 \pm 2.348\)**, **CD \(2.125 \pm 2.066\)**, and **HD \(11.078 \pm 9.112\)**. Removing sex also degrades performance to **MAE \(1.394 \pm 1.535\)**, **CD \(1.984 \pm 1.796\)**, and **HD \(10.534 \pm 7.947\)**. A **KCN-only** model without GCN gives **MAE \(1.592 \pm 3.614\)**, **CD \(2.178 \pm 2.546\)**, and **HD \(11.843 \pm 9.773\)**, while a **GCN-only** model without KCN gives **MAE \(2.042 \pm 3.363\)**, **CD \(2.740 \pm 2.302\)**, and **HD \(13.478 \pm 10.495\)**. Removing adversarial loss, even with both branches present, yields **MAE \(1.588 \pm 2.955\)**, **CD \(2.158 \pm 2.459\)**, and **HD \(11.248 \pm 9.258\)**. Using CD loss instead of \(L_1\), together with both branches and adversarial loss, yields **MAE \(1.515 \pm 2.100\)**, **CD \(2.357 \pm 1.627\)**, and **HD \(11.349 \pm 9.339\)** [2508.19862].

Qualitative examples are reported for follow-up generation at **3, 15, 39, and 45 months** from a baseline mesh. The predicted meshes are described as showing strong agreement with corresponding ground-truth follow-up meshes, with low CD maps and small diameter errors. The paper also states that MCMeshGAN better preserves **local deformations**, **smoother and more realistic surfaces**, and **overall anatomical plausibility**, while also acknowledging that **slight surface roughness** remains [2508.19862].

## 6. Temporal controllability, relation to prior mesh GANs, and limitations

A central property of MCMeshGAN is **temporal controllability**. Because the target interval is directly encoded and fed to the generator, a baseline mesh can be used to generate multiple predictions at different future or past times relative to the source timepoint. The interval can be positive for future prediction or negative for past reconstruction, and the paper explicitly frames this as support for both **retrospective** and **prospective** modeling. In the conditioning ablation, **age** is reported to have greater impact than **sex**, which the paper suggests is likely because age spans a larger continuous range, **35 to 93 years**, whereas sex is binary [2508.19862].

In the broader lineage of mesh-native adversarial generation, MCMeshGAN differs from earlier systems in both application domain and conditioning strategy. **MeshGAN** was introduced as the first intrinsic GAN architecture operating directly on 3D meshes for faces, using fixed-topology dense-correspondence face meshes, Chebyshev spectral graph convolutions, and a BEGAN autoencoder discriminator [1903.10384]. MCMeshGAN, by contrast, is described as **multimodal**, **conditional**, and **mesh-to-mesh**, with direct incorporation of **clinical metadata** and **time interval** for 3D aneurysm growth prediction. Relative to GAN frameworks that encode mesh geometry and texture as aligned UV-space signals and support class-, attribute-, and text-conditional generation of textured meshes from single-view image supervision, MCMeshGAN remains focused on registered anatomical surface meshes and clinical progression rather than geometry-texture synthesis [2006.07660].

The paper’s stated contributions are fourfold: the first **multimodal conditional mesh-to-mesh translation model** for 3D aneurysm growth prediction; a **dual-branch architecture** combining global **GCN** features and local **KCN** features; integration of **age**, **sex**, and **time interval** for personalized prediction; and introduction of the **TAAMesh** longitudinal 3D aneurysm mesh dataset [2508.19862].

Its stated limitations are equally specific. The model can still produce **rough mesh surfaces**; it was evaluated only on the authors’ **private dataset**; there is no public longitudinal aneurysm mesh benchmark for broader validation; **no external validation** is reported; and the framework has not yet been integrated into a **real clinical workflow**. The method also depends implicitly on strong topology assumptions because all meshes undergo non-rigid registration and are standardized to **10,000 vertices**. A plausible implication is that the model’s success depends not only on its generator design but also on the availability of high-quality longitudinal mesh registration and correspondence [2508.19862].

The source code for MCMeshGAN and the baseline methods is reported as publicly available at **https://github.com/ImperialCollegeLondon/MCMeshGAN**. The data themselves are described as curated from Hammersmith Hospital, but the text provided does not state that TAAMesh has been publicly released. In the paper’s own framing, MCMeshGAN represents a step toward **personalized 3D disease trajectory modeling**, with future work required on surface smoothness, generalization, and clinical deployment [2508.19862].

Source: https://www.emergentmind.com/topics/mcmeshgan