Observations of Plutino 2003 AZ84 suggest it could share similar properties to Orcus, or even Pluto’s satellite Charon. Because it is smaller, hence fainter, its surface composition could not be constrained from ground-based spectroscopy. The JWST/NIRSpec instrument was thus used to target 2003 AZ84, using the Fixed Slit (FS) mode, with both low and medium spectral resolutions, from 0.6 to 5.3 microns. To avoid any loss of spectral information, an optimal processing of these FS data is necessary. In this work, we propose such an optimal spectra extraction method, using an inverse problem approach. For each wavelength, the data acquired on the detector are considered as a 1D spatial Point Spread Function (PSF), with the amplitude giving the spectral value for this wavelength, plus some additive noise. A thorough model of the spatial PSF is thus essential to ensure the optimal processing. The spectrum is then extracted by minimizing the co-log-likelihood of the data, with an additive regularization term, to avoid noise amplification in the extraction. We present the results of our method for the spectra extraction of 2003 AZ84, and compare them to the spectra obtained with the JWST/NIRSpec pipeline, where the spectrum is extracted by integrating the flux in fixed size small aperture.