An Algorithm from Netflix Challenge to Speed Up Biological Imaging ?

Record-fast speeds make spontaneous Raman spectroscopy more practical for biomedical applications

Raman spectroscopy is a well-known non-invasive technique to determine the chemical composition of complex samples. For example, it has shown promise for identifying cancer cells and analyzing tissue in search of diseases. Although this method is remarkably simple (no sample preparation is needed), it typically requires image acquisition speeds that are too slow to capture the dynamics of biological specimens. Moreover, processing the massive amount of data generated by spectroscopic imaging is time-consuming, and often a limitation to study large area specimens.

An international collaboration in which the LPENS and the LKB are involved, developed a methodology to address two challenges simultaneously —increasing the acquisition speed and introducing a more straightforward way to acquire useful information from the spectroscopic images. Their study is published in the open-source optical journal Optica.

To speed up the imaging process, the researchers adapted their Raman system by replacing the expensive and slow cameras used in conventional setups with a cheap and fast spatial light modulator known as a digital micromirror device. This device selects groups of wavelengths that are detected by a highly sensitive single-pixel detector. They deliberately do not acquire the complete data set typically required for classical Raman spectroscopic imaging, in effect compressing the images as they are acquired. On the software side, they fill in the missing information exploiting a very special algorithm : it is actually a repurposed algorithm originally developed for Netflix’s 2009 movie preference prediction competition !

The researchers demonstrated their new methodology by imaging brain tissue and single cells, both of which exhibit high chemical complexity. Their results showed that the method can acquire images at speeds of a few tens of seconds, where traditional Raman would typically take minutes, concomitantly compressing the data — they have reduced the data size up to 64 times.

The researchers believe that the new approach should work with most biological specimens, but they plan to test it with more tissue types to demonstrate this experimentally. In addition to clinical tools, the method could be useful for biological applications such as algae characterization. They also want to improve the scanning speed of their system to accomplish sub-second image acquisition in the future. Such advance could make the simple, cheap, and label-free Raman imaging method practical for clinical applications such as tumor detection or tissue analysis.

Hilton B. de Aguiar, the coordinator of the collaboration, is a Junior Chair Research holder at the ENS Physics’ Department.

Original Paper : F. Soldevila, J. Dong, E. Tajahuerce, S. Gigan, H. B. De Aguiar, “Fast compressive Raman bio-imaging via matrix completion,” Optica 6, 341-346 (2019).

PNG - 157.6 ko

Figure 1 Compressive Raman Imaging - With the new compressive Raman approach, one can acquire less spectral data than traditionally required and then use the matrix completion algorithm to fill in information not recorded.

PNG - 441.8 ko

Figure 1 Raman microspectroscopy of opaque brain tissue --- Scale bar : 20 microns.

Credit : Hilton De Aguiar, École Normale Supérieure

Corresponding author :

Communication contact :