ANTSAI(1)

🎀 ANTSAI(1) — User Commands 🎀

✨ NAME ✨

antsAI 🦋 — part of the ANTS registration suite, bestie! It's giving precision alignment 💅

---

💖 DESCRIPTION 💖

👑 COMMAND: antsAI

okay so antsAI is the program that figures out the optimal linear transform parameters for lining up two images 🌸 It's literally the thing that makes your scans match up, no cap ✨

---

🫶 OPTIONS

• -d, --dimensionality 2/3 — okay this one forces the image to be treated as a specific-dimensional image (2D or 3D, you pick 🧚). If you don't specify it, antsAI tries to infer the dimensionality from the input image on its own. Main character energy 👑
• -m, --metric — the metric options, bestie 💜 Here's what's available:
  • Mattes[fixedImage,movingImage,<numberOfBins=32>,<samplingStrategy={None,Regular,Random}>,<samplingPercentage=[0,1]>]
  • GC[fixedImage,movingImage,<radius=NA>,<samplingStrategy={None,Regular,Random}>,<samplingPercentage=[0,1]>]
  • MI[fixedImage,movingImage,<numberOfBins=32>,<samplingStrategy={None,Regular,Random}>,<samplingPercentage=[0,1]>]

So Mattes is Mattes mutual information and it's the one they recommend 🔥 GC is global correlation, and MI is joint histogram mutual information 🎀

• -t, --transform — several transform options are giving variety 🦋 Pick from Rigid[gradientStep], Affine[gradientStep], Similarity[gradientStep], AlignGeometricCenters, or AlignCentersOfMass. For rigid, affine, and similarity, the gradientStep sets up the gradient descent optimization, and it gets scaled properly for each transform using the shift scales estimator. The other two just find the simple translation transform that aligns the specified image feature 🌈 Heads up, the images get read from the similarity metric option even though the metric isn't actually used 💀
• -p, --align-principal-axes — a boolean for aligning by principal axes ✨ The alternative is aligning with blobs (see -b) 🧚
• -b, --align-blobs numberOfBlobsToExtract — [numberOfBlobsToExtract,<numberOfBlobsToMatch=numberOfBlobsToExtract>] 🫧 Another boolean, this one for aligning by a set of blobs. The alternative is aligning by principal axes (see -p) 💜
• -s, --search-factor searchFactor — [searchFactor=20,<arcFraction=1.0>] 🌸 This is an incremental search factor, in degrees, that samples the arc fraction around the principal axis or the default axis 🦋
• -g, --translation-search-grid [stepSize=25, AxBxC=0x0x0] — a translation search grid in mm 🍓 It translates the moving image along each dimension in increments of the step size ✨
• -c, --convergence numberOfIterations — [numberOfIterations,<convergenceThreshold=1e-6>,<convergenceWindowSize=10>] 🔥 This is the number of iterations, bestie 💅
• -x, --masks fixedImageMask — [fixedImageMask,movingImageMask] 🎀 Image masks that limit which voxels the metric considers 🫶
• -o, --output outputFileName — specify the output transform, which comes out as an ITK .mat file 💖
• --random-seed seedValue — use a fixed seed for random number generation 🌈 This overwrites the default fixed seed. It can be any nonzero int value, but if you pass zero, the system time gets used instead ⏰
• -v, --verbose (0)/1 — verbose output, so you can see all the tea ☕💅
• -h — prints the short version of the help menu 🙈
• --help — prints the help menu 💜 <VALUES>: 1 🌸

---

🦋 SIGN-OFF 🦋

That's the whole tea, bestie! This page came from antsAI 2.6.4+dfsg, dated December 2025, and I'm obsessed with how iconic it is 💖✨ Stay aligned, stay slay 🫶💅