System for determining the presence of a substance of interest in a sample

Inventors

Troy, Jerome • Edwards, John W. • Goldman, Heather • Joiner, David • Miller, William

Assignees

Smiths Detection Inc

Abstract

A detection device for detecting the presence of a substance of interest in a sample is described. The device can include a data store comprising executable instructions for at least one convolutional neural network, CNN, configured to process images: and a processor coupled to the data store and configured to execute the instructions to operate the at least one CNN. The detection device can be configured to: obtain spectrometry data, operate a first one of the CNNs to process the spectrometry data to obtain a first CNN output; apply a mask to the spectrometry data to obtain masked data; operate a second one of the CNNs to process the masked data to obtain a second CNN output; and determine if the substance of interest is present in the sample based on both the first CNN output and the second CNN output.

Core Innovation

The invention is a detection system and detection device for detecting the presence of a substance of interest in a sample using spectrometry data represented as at least one two dimensional array. Spectrometry data is generated by a detector or obtained for a device, and the system operates at least one convolutional neural network, CNN, configured to process images. The system uses a first CNN to process the spectrometry data and obtain a first CNN output.

The system then applies a mask to the spectrometry data to obtain masked data, and operates a second CNN to process the masked data and obtain a second CNN output. The processor determines if the substance of interest is present in the sample based on both the first CNN output and the second CNN output. The decision is grounded in outputs from the CNNs operating on both unmasked and masked representations.

The described approach is implemented across example system and device architectures and supports multiple spectrometry modalities, including mass spectrometry, Raman spectroscopy, optical spectrometry, Fourier Transform Infrared, ion mobility spectrometry, radiological spectrometry, and biological agent spectrometry. The document further describes converting spectrometry array data into 2D array image representations, including raster image generation concepts, and determining presence using probability and threshold decision rules, with optional ensemble and voting across multiple CNN outputs and masks.

The document additionally describes training and update behavior tied to known outcomes, including updating CNN instructions when labeled ground truth contradicts outputs. It describes that masking can be produced by reducing or zeroing intensity in a region of the spectrometry data. Overall, the core is using two CNN outputs from original and masked two-dimensional array image representations of spectrometry data to decide presence of the substance of interest.

Claims Coverage

The document includes two independent claims, a detection system and a detection device. Each independent claim contains the same core set of inventive features: two CNN processing paths, unmasked and masked spectrometry data arranged in two-dimensional arrays, whose outputs are jointly used to determine presence.

Across the independent claims, the inventive coverage centers on arranging spectrometry data in at least one two-dimensional array, processing the unmasked array with a first CNN, processing a masked version with a second CNN, and determining the presence of the substance of interest based on both CNN outputs.

Stated Advantages

Documented Applications