Journal: Scientific Data

Article Title: SLIMBRAIN database: A multimodal image database of in vivo human brains for tumour detection

doi: 10.1038/s41597-025-04993-y

Figure Lengend Snippet: Schematic of the calibration library procedure followed in the operating room with the simplified 3D model of the SLIMBRAIN prototype 3. ( a ) Illustration of the positioning of the acquisition system with respect to the diffuse reflectance target. ( b ) White and dark references obtained with the Ximea snapshot and Headwall linescan cameras.

Article Snippet: Although three HS cameras have been utilized to acquire images, the images taken from the Ximea linescan camera presented many difficulties when reconstructing HS cubes with the nonflat surface of the brain.

Techniques:

Journal: Scientific Data

Article Title: SLIMBRAIN database: A multimodal image database of in vivo human brains for tumour detection

doi: 10.1038/s41597-025-04993-y

Figure Lengend Snippet: Representation of the spatial and spectral information captured in every line with the Headwall linescan HS camera. The information is saved using the BIL binary format with a header file indicating how to read the data.

Article Snippet: Although three HS cameras have been utilized to acquire images, the images taken from the Ximea linescan camera presented many difficulties when reconstructing HS cubes with the nonflat surface of the brain.

Techniques:

Journal: Scientific Data

Article Title: SLIMBRAIN database: A multimodal image database of in vivo human brains for tumour detection

doi: 10.1038/s41597-025-04993-y

Figure Lengend Snippet: Normalized reflectance spectral responses with correlation values when illuminating the polymer reference with lamps B and D from Fig. and captured with different sensors. Spectrometer measurements were performed with a fibre optic orthogonal to the polymer reference from a 5 cm distance. The hyperspectral camera captures a scene of the polymer at approximately 72.5 degrees and 40.5 cm distance and uses a region of interest of 25 × 25 pixels. Note that captures taken at 90 ∘ and 0 ∘ would mean pointing the camera to a vertical wall and to the floor, respectively. These responses from the HS cameras are the mean spectral signatures of the 25 × 25 pixel polymer pixels with their corresponding standard deviations, shown with shaded colours between their corresponding means. The spectrometer uses 2055 bands and covers the 350–925 nm range. The HS linescan VNIR camera covers the 400–1000 nm spectral range and uses 369 bands. Finally, the HS snapshot NIR camera spectra ranges from 660–950 nm and includes 25 bands. The Pearson correlation is presented after the measured bands of each sensor are compared with the polymer reference response. The visible (VIS) spectra is illustrated with rainbow shading in the 380–740 nm range.

Article Snippet: Although three HS cameras have been utilized to acquire images, the images taken from the Ximea linescan camera presented many difficulties when reconstructing HS cubes with the nonflat surface of the brain.

Techniques: Polymer

Journal: Scientific Data

Article Title: SLIMBRAIN database: A multimodal image database of in vivo human brains for tumour detection

doi: 10.1038/s41597-025-04993-y

Figure Lengend Snippet: Raincloud plot with the SAM thresholds used for every reference label pixel. Obtained during the ground-truth labelling procedure for the Headwall linescan brain captures with IDs, which can be found inside the zipped folder “PaperExperiments” on the e-cienciaDatos repository, as well as in the Headwall_Linescan_GT_Patient_IDs.txt file.

Article Snippet: Although three HS cameras have been utilized to acquire images, the images taken from the Ximea linescan camera presented many difficulties when reconstructing HS cubes with the nonflat surface of the brain.

Techniques:

Journal: Scientific Data

Article Title: SLIMBRAIN database: A multimodal image database of in vivo human brains for tumour detection

doi: 10.1038/s41597-025-04993-y

Figure Lengend Snippet: ROC curves obtained on the test set with two RF models trained with labelled pixels from both the snapshot and the linescan HS cameras independently. Each class contains the AUC score in the legend expressed in %.

Article Snippet: Although three HS cameras have been utilized to acquire images, the images taken from the Ximea linescan camera presented many difficulties when reconstructing HS cubes with the nonflat surface of the brain.

Techniques: