When AI help us to monitor babies

https://theconversation.com/a-camera-combined-with-ai-could-help-further-diagnosis-of-movement-disorders-like-spinal-muscular-atrophy-in-newborns-293152

Some interesting videos about my work.

 

Red MP3 Reader

Well, Red is awesome. The idea here was to combine macOS tools like afplay with a user-friendly GUI. It's done!

Face Update

 After a few hours of work and discussions with Red/Sensei (https://ask.lang.red/chat), the Face software—which I wrote a few years ago to measure facial bone growth in children—is now available for macOS-ARM 64-bit.  We have two versions (https://github.com/ldci/Face)—one for 32-bit and one for 64-bit!

OCR to TTS

With macOS on ARM-64, Tesseract and Say work together. The OCR reads the image and converts it into text. Then the TTS reads and speaks the text.

(see https://github.com/ldci/Red_KIS/tree/main/Tesseract)

Red 64-bit

 Thanks to Qtxie's superb work, Red now runs in 64-bit mode on Apple's ARM processors. I'm enjoying programming with Red again!

Rebol R3 skeleton

Using erode and dilate is similar to OpenCV morphologyEx (not supported by Rebol 3) 

#!/usr/local/bin/r3
Rebol [
]
;--all images are binary with 1 channel
;--using erode and dilate is similar to OpenCV morphologyEx (not supported by Rebol 3)
cv: import opencv
with cv [
    img: imread/with "../images/char.png" IMREAD_GRAYSCALE ;--source image as GS
    threshold :img :img 127 255 THRESH_BINARY ;--binary thresholding
    namedWindow win1: "Source"
    moveWindow win1 0x0
    imshow/name img "Source"
    skel: Matrix [:img/size CV_8UC1] ;--1 channel matrix (8-bit)
    element: getStructuringElement MORPH_CROSS 3x3 -1x-1 ;--structuring element for the kernel
    until [
    eroded: erode :img none element -1x-1 1 ;--erode
    temp: dilate :eroded none element -1x-1 1 ;--then dilate
    subtract :img :temp :temp ;--img source - eroded image
    bitwise-or :skel :temp :skel ;--just OR operator
    img: :eroded ;--update source image
    maxi: second minMaxLoc img ;--get maxi value
    maxi = 0 ;--repeat until maxi = 0
    ]
    namedWindow win2: "Skeleton"
    moveWindow win2 250x0
    imshow/name skel win2
    waitKey 0
    destroyAllWindows
]

Heat diffusion

From https://en.wikipedia.org/wiki/Heat_equation

"In mathematics and physics (more specifically thermodynamics), the heat equation is a parabolic partial differential equation. The theory of the heat equation was first developed by Joseph Fourier in 1822 for the purpose of modeling how a quantity such as heat diffuses through a given region. Since then, the heat equation and its variants have been found to be fundamental in many parts of both pure and applied mathematics". 

The idea is to see how Rebol or Red can solve this problem. Let’s imagine a coin heated at its center at T0.0 and calculate the temperature distribution toward the edges of the coin. We get a nice Gaussian curve.

Now, let’s look at how the phenomenon changes over time (from 0.0 to 0.10). We still see nice Gaussian curves whose mean decreases over time. We can also model the cooling of the metal part. 

Due to vector datatype  (https://github.com/Oldes/Rebol3) these equations are very easy to solve.

#!/usr/local/bin/r3

REBOL [

]

b2d: import 'blend2d ;--use blend2d (draw module)

cv2: import 'opencv ;--use OpenCV for visualisation

;--Global Physical Parameters

k: 0.5 ;--Diffusion coefficient (Thermal diffusivity coefficient of material)

l: 1.0     ;--Domain size

itime: 0.1   ;--Integration time

;--Global Numerical Parameters

nx: 100     ;--Number of grid points

nt: 1000   ;--Number of time steps

dx: l / (nx - 1)   ;--Grid step (Spatial step size: 0.0101010101010101)

dt: itime / nt   ;--Grid step (Time step size :0.0001)

;--Based on Oldes's code

linSpace: func [

    "Generates n linearly spaced numbers from a to b (inclusive)."

    a [number!]  "Start"

    b [number!]  "End"

    n [integer!] "Number of samples"

    /no-end      "Don't include end value in the result"

    /local vec div step i ;--local values

][

    vec: make vector! [f64! :n]

    div: either no-end [n][n - 1]

    step: (b - a) / div

    repeat i n [vec/:i: a + (step * (i - 1))]

    vec

]

;--Initialisation (using global variables)

init: does [

x: linSpace 0.0 1.0 nx ;--linear spacing between 0.0 and 1.0

t: (x * 2.0 * pi) ;--t is a vector

repeat i nx [t/:i: sin t/:i] ;--in radians

;--use blend2d commands in plot block

plot: compose [

font %/System/Library/Fonts/Geneva.ttf

fill black

text 0x10  10 "+1.0"

text 0x204 10 "0"

text 0x390 10 "-1.0"

pen black

line 5x200 410x200 

]

]

;--Heat diffusion equation (using global variables)

diffusion: does [

vect: make vector! [f64! :nx]

posy: 0

n: 0

while [n <= nt] [

j: 2

while [j < nx] [

vect/:j: dt * k * (t/(j - 1) - (2.0 * t/:j) + t/(j + 1)) / (dx ** 2)

++ j

]

repeat j nx [t/:j: t/:j + vect/:j]

;--Plot every 100 time steps

if zero? n % 100 [ 

acolor: random white

posy: posy + 15

either n = 0 [tt: 0.0] [tt: round/to n % (n * dt) 0.001]

tt: ajoin ["time: " tt]  

pos: as-pair 340 posy

append plot reduce ['fill (acolor) 'text (pos) 12 (tt) 'pen (acolor) 'line]

xx: 2

foreach val t [

append plot as-pair xx * 4 200 - (val * 195) 

++ xx 

]

]

++ n

]

]

;************************ Main Program ************************

init 

diffusion

img: b2d/draw 410x400 :plot

img: cv2/resize img 150% 

cv2/imshow/name :img "Heat Diffusion"

print "Any key to close"

cv2/waitKey 0

Using Filter2D with Rebol-OpenCV

 Rebol-OpenCV (https://github.com/Oldes/Rebol-OpenCV) Filter2D is really fantastic for creating various filters on image.

filter2D: [
"Convolves an image with the kernel."
src    [handle!] "cvMat"
dst    [handle! none!] "cvMat"
ddepth [integer!] "desired depth of the destination image"
kernel [handle!] "convolution kernel (or rather a correlation kernel), a single-channel floating point matrix"
anchor [pair! integer!] "position of the anchor within the element"
delta  [number!] "value added to the filtered pixels before storing them in dst"
/border "border mode used to extrapolate pixels outside of the image"
type [integer!] "one of: [0 1 2 4 5 16]"
]

This is an example of creating a Kuwahara_filter

#!/usr/local/bin/r3
Rebol [
]
;--https://en.wikipedia.org/wiki/Kuwahara_filter
cv: import opencv
with cv [
    src: imread/with "../images/mandrill.jpg" -1 ;--(IMREAD_UNCHANGED)
    dst: Matrix :src ;--create a matrix
    ;--using Rebol vector type for a 3x3 kernel
    vec: #(f32! [-0.5 1.5 -0.5 1.5 -3.0 1.5 -0.5 1.5 -0.5]) ;--sum = 1
    filter: Matrix [CV_32FC1 3x3 :vec] ;--a 3x3 kernel
    filter2D :src :dst -1 :filter -1x1 0 ;-1 same image depth
    ;--show result
    imshow/name src "Source"
    imshow/name dst "Kuwahara"
    moveWindow "Source"  0x0
    moveWindow "Kuwahara" 260x0
    waitKey 0
    destroyAllWindows
]

The trick is to use vector datatype implemented by Oldes to create the kernel used by the filter2D function.

The result for the Kuwahara_filter:

You can find here https://github.com/ldci/R3_OpenCV_Samples/tree/main/image_filtering a lot of examples for generating filters on image.

A simple GUI for Rebol3

Rebol3, developed by Oldes (https://github.com/Oldes/Rebol3), is improving day by day. Unfortunately, Rebol3 does not yet have a VID (except for Windows as a proof of concept), which means that attractive interfaces cannot yet be created. 

Fortunately, Oldes added a nice modules management to Rebol3, which allows this problem to be overcome. 

In this code, we use two modules:

Blend2D (https://github.com/Siskin-framework/Rebol-Blend2D) 

OpenCV (https://github.com/Oldes/Rebol-OpenCV) 

The idea is as follows: we use the Blend2D module to create the interface for our application and the OpenCV module to display the result. 

With Blend2D module, I created five buttons (simple blend2D boxes). Each button is associated with a Rebol3 function that will execute when the button is clicked. Button 1 calls the loadImage function, which allows you to select an image and display it. Buttons 2 and 3 call the convertTo function, which displays the image in grayscale or in HSV color space. Button 4 displays the original image and button 5 exits the application.

Now, the question is how to associate mouse movements with different buttons. Blend2D does not have a mouse manager, which is where the OpenCV module comes in.

Oldes has introduced a mouse event handler into the OpenCV module for these events:

• MOUSEMOVE
• LBUTTONDOWN
• RBUTTONDOWN
• MBUTTONDOWN
• LBUTTONUP
• RBUTTONUP
• MBUTTONUP
• LBUTTONDBLCLK
• RBUTTONDBLCLK
• MBUTTONDBLCLK
• MOUSEWHEEL
• MOUSEHWHEEL

With the cv/setMouseCallback function, we can retrieve the event type, the x and y position of the mouse, and the event flag. We also obtain the mouse position as a pair! x y.

;--OpenCV mouse callback in context

ctx: context [

    on-mouse-click: func [

        type  [integer!]

        x     [integer!]

        y     [integer!]

        flags [integer!]

    ][

    pos: mcb/pos ;--mouse position as a pair!

        if type == cv/EVENT_LBUTTONDOWN [

        if pos/y < 20 [

        case [

        all [pos > 0x0 pos < 80x20]   [loadImage]         ;--button 1

        all [pos > 80x0 pos < 160x20] [if isLoaded? [convertTo/GS fileName]]        ;--button 2

        all [pos > 160x0 pos < 240x20][if isLoaded? [convertTo/HSV fileName]]    ;--button 3

        all [pos > 240x0 pos < 320x20][showSource]    ;--button 4

        all [pos > 320x0 pos < 400x20][quit] ;--button 5

        ]

        ]

        ]

    ]

]

In the on-mouse-click function, we first check that the mouse is within the button area. Then we check that the mouse is on the selected button. If so, we execute the function associated with the button. In this example button 3 is clicked and convertTo function is executed.

Added new faces for Rebol3 GUI

CheckBox

RadioBox

Horizontal Scroller

Toggle On

Progress

The code is here: https://github.com/ldci/R3_tests/tree/main/GUI

Virginia Project: A first paper from R2P2 Lab

https://www.frontiersin.org/journals/pediatrics/articles/10.3389/fped.2025.1636667/full

First, the Virginia software, written with Red language, isolated neonatal anatomy through PyTorch-based PointRend segmentation combined with morphological filtering.

Second, radiometric decoding via ExifTool and ImageMagick extracted pixel-level temperature values mapped to anatomical regions of interest (chest, extremities). Finally, quantitative thermal metrics were derived, including median body surface, temperature and spatial thermal variability (interquartile range).

A key advantage of this automated pipeline is its low operator dependence; once the image is acquired, the entire segmentation and feature extraction process is software-driven, minimizing human interpretation bias.

BlurHash with r3

 

What is BlurHash? (from https://uploadcare.com/blog/blurhash-images/ and https://github.com/woltapp/blurhash )

BlurHash is a lightweight way to represent a blurred version of an image that was invented by the Wolt team. It is represented by a short piece of text that when decoded can produce a low quality version of an image which can be shown to the user while the actual image is being loaded.

The idea behind BlurHash is to design a placeholder that is close to the original image but has a smaller size. This makes it possible to send the placeholder to the client side of your application, which reduces the time that the user spends waiting for the page to load.

Oldes wrote an extension for Rebol 3 that makes it very easy to use BlurHash.

See https://github.com/Siskin-framework/Rebol-BlurHash

#!/usr/local/bin/r3
Rebol [
    title: "Rebol/BlurHash test"
]
blurhash: import 'blurhash
cv:         import 'opencv

image: load %../pictures/test1.tiff ;--use your own image
print ["Encoding image of size" as-yellow image/size]
hash: blurhash/encode image
print ["String: " hash]
print ["Decoding hash into image"]
blured: resize blurhash/decode hash 32x32 image/size

;--use OpenCV extension for visualisation
with cv [
print "Source Image"
imshow/name image "Source"
waitkey 0
print "Blured Image"
imshow/name blured "Blured"
waitkey 0
print "Any key to close"
]

Thanks to Oldes!😀

SEDFormer

Exciting news! Our paper "SEDformer: Path Signatures and Transformers to Predict Newborns Movement Symmetry" has been accepted at the International Joint Workshop of Artificial Intelligence for Healthcare (HC@AIxIA) and HYbrid Models for Coupling Deductive and Inductive ReAsoning (HYDRA), ECAI 2025.

This research introduces SEDformer, a FEDformer variant that integrates path signatures to enhance newborn movement symmetry prediction - a crucial step toward automated early screening tools for motor development disorders.

Thank you to the entire team for this interdisciplinary collaboration between applied mathematics, AI, and pediatric medicine.

 Rambaud, P., Rimmel, A., Trabelsi, I., Zini, J., Wodecki, A., Motte Signoret, E., Jouen, F., Tomasik, J., Bergounioux, J. (2025). SEDformer: Path Signatures and Transformers to Predict Newborns Movement Symmetry. International Joint Workshop of Artificial Intelligence for Healthcare (HC@AIxIA) and HYbrid Models for Coupling Deductive and Inductive ReAsoning (HYDRA)" ECAI 2025.

Nonlinear acoustic phenomena tune the adults’ facial thermal response to baby cries with the cry amplitude envelope

A new article on the use of thermography: https://royalsocietypublishing.org/doi/10.1098/rsif.2025.0150.

I am very proud of this innovative article. The first thermal image processing codes used Red and Rebol-3.

Yolo CDD

Our latest article using YOLO is here: https://ieeexplore.ieee.org/document/11078264.

Using FLIR cameras for research

The IR cameras from FLIR (https://www.flir.fr) are little marvels of technology that can acquire quality IR images. What I like about FLIR is that the data format remains the same regardless of the camera used. For my work in neonatal medicine, I use either the C3 model (basic) or the 650SC model (much more expensive and more powerful).

FLIR generates four types of image. The first is the IR image, whose resolution varies from 320x240 to 640x480 pixels, depending on the camera model. The second is an RGB image, up to six times larger than the IR image. The third image, which can be the same size as the IR image or smaller (80x60 pixels).  It contains temperatures in degrees Celsius. Finally, the last image is the color palette of the IR image. So you can imagine all the calculations that have to be made to obtain comparable images. You'll find various toolkits in Python, MatLab, R... that allow you to process these different images. Unfortunately, these libraries are not universal and often depend on other libraries that are not easy to install.

That's why, as part of the Virginia project (https://uniter2p2.fr/projets/), I designed an easy-to-use FLIR image processing module for the Red and Rebol 3 languages.

THE FLIR MODULE

This module has been tested with various FLIR cameras. Its main function is to decode the metadata contained in a radiometric file and extract the visible image (RGB), the infrared image (IR), the color palette associated with the IR image and the temperatures associated with each pixel.

This module calls on two external programs that are installed by default on macOS and Linux.

ExifTool (https://exiftool.org), written and maintained by Phil Harvey, is a fabulous program written in Perl that lets you read and write the metadata of a large number of computer files. ExifTool supports FLIR files. It runs on all MacOs, Linux and Windows platforms.

ImageMagick (https://imagemagick.org/index.php) is an open-source software package comprising a library and a set of command-line utilities for creating, converting, modifying and displaying images in a wide range of formats. The FLIR module essentially uses the magick utility for MacOs and Linux versions. For Windows, use a portable version that supports 16-bit images (https://imagemagick.org/archive/binaries/ImageMagick-7.1.0-60-portable-Q16-x64.zip) and the magick command.

The module calls for:

rcvGetFlirMetaData: This function takes the name of the FLIR file as a parameter (in the form of a character string). It returns all the information in the patient's irtmp/exif.red file in a format that can be directly processed by Red or Rebol 3.

rcvGetVisibleImage: This function extracts the RGB image from the FLIR file and saves it in the irtmp/rgb.jpg file.

rcvGetFlirPalette: Extracts the color palette contained in the FLIR file and samples it for a linear range of values [0..255]. The extracted image is saved as irtmp/palette.png.

rcvMakeRedPalette: Exports the color palette as a block for fast processing with Red or Rebol 3.

rcvGetFlirRawData: Extracts raw temperature data (in 16-bit format) into the irtmp/rawimg.png file.

rcvGetPlanckValues: Retrieves all constants required for accurate temperature calculations.

rcvGetImageTemperatures: This function uses the previous two functions to calculate the temperature of each image pixel as an integer value. It creates the image tmp/celsius.pgm. This is a 16-bit image with a maximum value of 65535. It's a simple text file containing the image size and the 16-bit values of each pixel.

rcvGetTemperatureAsBlock : The temperatures contained in the irtmp/celsius.pgm image are returned as a real value (e.g. 37.2) in the block passed as a parameter to the function. This is a dynamic calculation.

WHY IMAGES ALIGNMENT IS FUNDAMENTAL?

The neural networks we use to identify babies' bodies have not been trained on thermal images, which are difficult to process, but work very well with RGB images. Once the baby's body is correctly identified in the RGB image, we can use the resulting body mask to retrieve the temperatures in the thermal image.  Obviously, we can't use the RGB image directly, but the RGB image aligned with the thermal image. 

In previous versions of Virginia, I wrote a rather complicated algorithm for aligning thermal and IR images. Studying the code, I found that it was possible to make it simpler. There are three values that will help us: Real2IR, offsetX and offsetY, which come from the rcvGetFlirMetaData function. Real2IR allows us to calculate the ratio between the RGB image and the thermal image.  OffsetX and offsetY are the X and Y offset coordinates to be applied to find the origins of the ROI in the RGB image. If these values are equal to 0, alignment is not required.

The result is perfect!

The code for Rebol 3 is here:

https://github.com/ldci/FLIR

The code for Red is here:

https://github.com/ldci/redCV/blob/master/samples/image_thermal/Flir/align.red

Gnuplot

I really like Gnuplot (http://www.gnuplot.info), a command-line utility for creating sophisticated graphics. It's in line with Red and Rebol's philosophy: Keep It Simple (KIS). Here's an example:

#!/usr/local/bin/gnuplot -persist
set hidden3d
set isosamples 50,50
set ticslevel 0
set pm3d
set palette defined (0 "black", 0.25 "blue", 0.5 "green", 0.75 "yellow", 1 "red")
splot sin(sqrt(x**2+y**2))/sqrt(x**2+y**2)

And the result:

Just in a few lines of code. Great!
 

Statistics on image

With Red or Rebol R3, the vector! type is ideal for fast numerical calculations. 

Recently, Oldes has introduced new properties for vectors in R3 that allow you to obtain the descriptive statistics of a vector in one basic step. Great work!

An example

#!/usr/local/bin/r3
REBOL [ 
]
vect: #(float64! [1.62 1.72 1.64 1.7 1.78 1.64 1.65 1.64 1.66 1.74])
print query vect object!

The result:

signed: #(true)

type: decimal!

size: 64

length: 10

minimum: 1.62

maximum: 1.78

range: 0.16

sum: 16.79

mean: 1.679

median: 1.655

variance: 0.02529

population-deviation: 0.0502891638427206

sample-deviation: 0.0530094331227943

But this can also be applied to images!

An example 

#!/usr/local/bin/r3

REBOL [ 

]

cv: import 'opencv

with cv [

filename: %../images/lena.png         ; --use your own image

mat: imread/with filename 2 ;--read as grayscale image with one channel

imshow/name mat filename ;--display the image  with file name as title

moveWindow filename 200x10 ;--move window

vect: get-property mat MAT_VECTOR ;--get matrix values as a vector    

print query vect object!

print "A key to quit"

waitKey 0

]

The result:

signed: #(false)

type: integer!

size: 8

length: 65536

minimum: 2

maximum: 225

range: 223

sum: 4377641

mean: 66.7975006103516

median: 64.0

variance: 126148517.630557

population-deviation: 43.87338169177

sample-deviation: 43.873716422939

Efficient :)

 

K-means algorithm

 The K-means algorithm is a well-known unsupervised algorithm for clustering that can be used for data analysis, image segmentation, semi-supervised learning... The k-means clustering algorithm is an exclusive method: a data point can exist in only one cluster.

K-means is an iterative centroid-based clustering algorithm that partitions a dataset into similar groups based on the distance between their centroids. The centroid (or cluster center) is either the mean or the median of all points.

Given a set of points and an integer k, the algorithm aims to divide the points into k groups, called clusters, that are homogeneous.

In this sample we generate a set of aleatory points in an image.

For processing data, we create a Red/Rebol object such as 

;--an object for storing values (points and clusters)

point: object [

x: 0.0 ;--x position

y: 0.0 ;--y position

group: 0 ;--cluster number (label)

]

The first step is to randomly define k centroids and associate them with k labels. Then, for each point, we calculate x and y Euclidian distances to the centroids and associate the point with the closest centroid and its corresponding label. This labels our data.

Secondly, we recalculate centroids, which will be the center of gravity of each labeled cluster of points. We repeat these steps until a convergence criterion is reached: centroids no longer move from the previous ones.

You will find the documented code for Red and Rebol 3 here:

 https://github.com/ldci/R3_OpenCV_Samples/tree/main/image_kmeans

Compress and Uncompress Images

A few years ago, I presented a way of compressing images with the Red zlib proposed by Bruno Anselme (https://redlcv.blogspot.com/2018/01/image-compression-with-red.html). Since then, Red and Oldes's Rebol 3 have implemented different compression methods that are faster and simpler to use. 

Both languages feature a compress function. Input data can be string or binary values, which is useful for RGB images. Returned values are binary. Both languages use lossless compression methods. 

Red and R3 share the following methods: 

deflate: A lossless data compression format that combines the LZ77 algorithm with Huffman coding.

zlib: Implements the deflate compression algorithm and can create files in gzip format. This library is widely used, due to its small size, efficiency and flexibility.

gzip: gzip is based on the deflate algorithm.

R3 adds a few more algorithms: 

br: Compression Brotli. A fast alternative to GZIP compression proposed by Google.

crush: A lossless compression package developed by the NASA.

lzma: Lempel-Ziv-Markov chain algorithm, is a lossless data compression algorithm.

As these methods are variations on deflate compression, the compression ratio doesn't vary much from one method to another. The difference is in the speed of compression.

 

Of course, both languages have a decompress function. Input data is binary, and the method used must be the same as that chosen for compression.   

Here's a minimalist example of code for Red and R3.  

method: 'zlib ;--a word

img: load %../pictures/in.png         ;--use your own image

bin: img/rgb ;--image as RGB binary

print ["Method    :" form method]

print ["Image size:" img/size]

print ["Before compression:" nU: length? bin]

t: dt [cImg: compress bin method]         ;--R3/Red compress

print ["After  compression:" nC: length? cImg]

ratio: round/to 1.0 - (nC / nU) * 100 0.01                 ;--compression ratio

print ["Compression :" form ratio "%"]

print ["Compress    :" third t * 1000  "ms"]                 ;--in msec

t: dt [uImg: decompress cImg method]         ;--R3/Red decompress

print ["Decompress  :" third t * 1000  "ms"]                 ;--in msec

print ["After decompression:" length? uImg]

The result:

Method    : zlib

Image size: 1920x1280

Before compression: 7372800

After  compression: 4011092

Compression : 45.6 %

Compress    : 46.298 ms

Decompress  : 26.706 ms

After decompression: 7372800

Fast and efficient!