Add kmeans module

.devcontainer/bash_history.sh

@@ -0,0 +1,4 @@
+#!/usr/bin/env bash
+
+SNIPPET="export PROMPT_COMMAND='history -a' && export HISTFILE=/commandhistory/.bash_history"
+echo "$SNIPPET" >> "${HOME:-/root}/.bashrc"

.devcontainer/devcontainer.json

@@ -0,0 +1,41 @@
+{
+    "name": "Beyond FA devcontainer - SCIL team",
+    "build": {
+        "dockerfile": "../Dockerfile",
+        "args": {
+            "BASE_IMAGE": "scilus/scilus:latest"
+        }
+    },
+    "onCreateCommand": "bash .devcontainer/bash_history.sh",
+    "postStartCommand": "git config --global --add safe.directory ${containerWorkspaceFolder}",
+    "features": {
+        "ghcr.io/devcontainers/features/git:1": {},
+        "ghcr.io/devcontainers/features/docker-in-docker:2": {
+            "dockerDashComposeVersion": "none",
+            "installDockerComposeSwitch": false
+        },
+        "ghcr.io/robsyme/features/nextflow:1": {}
+    },
+    "remoteUser": "root",
+    "mounts": [
+        {
+            "target": "/tmp",
+            "type": "volume"
+        },
+        {
+            "source": "nf-neuro-bash-history",
+            "target": "/commandhistory",
+            "type": "volume"
+        }
+    ],
+    "customizations": {
+        "vscode": {
+            "extensions": [
+                "KorbinianEckstein.niivue",
+                "nextflow.nextflow"
+            ]
+        }
+    },
+    "init": true,
+    "privileged": true
+}

.gitignore

@@ -0,0 +1,4 @@
+work/
+.nextflow.log*
+.nextflow/
+results/

Dockerfile

@@ -0,0 +1,7 @@
+ARG BASE_IMAGE=scilus/scilus:latest
+FROM $BASE_IMAGE
+
+RUN apt-get update && apt-get -y install \
+        git \
+        openjdk-17-jre \
+    && rm -rf /var/lib/apt/lists/*

modules/local/kmeans/main.nf

@@ -0,0 +1,59 @@
+
+process KMEANS {
+
+container "scilus/scilus:latest"
+
+input:
+    tuple val(meta), path(image)
+output:
+    tuple val(meta), path("labels/*.nii.gz"), emit: labels
+    tuple val(meta), path("*_centroids.npy"), emit: centroids
+
+script:
+    def prefix = task.ext.prefix ?: "$meta.id"
+    // K-means parameters
+    def nb_regions = task.ext.nb_regions
+    def algorithm = task.ext.algorithm ?: "lloyd"
+    def init_method = task.ext.init_method ?: "k-means++"
+    def nb_init = task.ext.nb_init ?: "auto"
+    def max_iter = task.ext.max_iter ?: 300
+    def norm_tolerance = task.ext.norm_tolerance ?: 1e-4
+    def seed = task.ext.seed ?: 1234
+    // Extract labels parameters
+    def background_label = task.ext.background_label ? "--background $task.ext.background_label" : ""
+    """
+    python3 - << ENDSCRIPT
+
+    import nibabel as nib
+    import numpy as np
+    from sklearn.cluster import KMeans
+
+    bin_image = nib.load("$image")
+    positions = np.nonzero(bin_image.get_fdata().astype(bool))
+    positions = np.asarray(positions).reshape((3, -1)).T
+
+    kmeans = KMeans(
+        n_clusters=$nb_regions,
+        init="$init_method",
+        n_init=${nb_init instanceof java.lang.String ? "\"" + nb_init + "\"" : nb_init},
+        max_iter=$max_iter,
+        tol=$norm_tolerance,
+        random_state=$seed,
+        algorithm="$algorithm"
+    ).fit(positions)
+
+    output = np.zeros_like(bin_image.get_fdata(), dtype=np.int16)
+    output[positions.T[0], positions.T[1], positions.T[2]] = kmeans.labels_ + 1
+
+    nib.save(nib.Nifti1Image(output, bin_image.affine), "binary_as_labels.nii.gz")
+    np.save(f"${prefix}_centroids.npy", kmeans.cluster_centers_)
+
+    ENDSCRIPT
+
+    scil_labels_split_volume_by_ids.py binary_as_labels.nii.gz \
+        $background_label \
+        --out_dir labels \
+        --out_prefix $prefix
+    """
+
+}