Single-image based methods used for non-invasive volume estimation of cancer spheroids: a practical assessing...

Cancer multicellular spheroids are commonly used as 3D tumour models for testing drugs and radiotherapy treatments. The volume plays a key role in analysis of the results. Several methods have been proposed in the literature to compute the spheroid's volume from one 2D microscopy image (i.e. a single projection). However, the literature lacks reviews summarising the different methods available. Furthermore, there are no well-established approaches by which to compare the different methods and determine the best one.

In this work we (a) revise the existing single-image based methods used to estimate the volume of multicellular spheroids, also providing different implementations for classical spherical and ellipsoidal pre-defined models; (b) present an upgrade of a volume estimation software recently proposed, Reconstruction and Visualization from a Single Projection (ReViSP), just validated by using four real spheroids imaged in 3D with a light-sheet microscope; (c) propose a quality assessing approach for single-image based methods, relying on 3D home-made macroscopic synthetic models mimicking the shapes of real multicellular spheroids.

Seven image-based methods used to estimate the volume of spheroids were compared using six 3D home-made synthetic models. First, the material used to make the synthetic models was characterised to estimate its density. Then, the ground-truth volume of the 3D models was measured by simply weighing them. The volume instances estimated by the different methods were compared with ground truth. ReViSP attained the best result three times out of six and on average.

The results obtained proved that (a) different implementations for the classical spherical and ellipsoidal pre-defined models may lead to very different results; (b) ReViSP is the best single-image based method available today to estimate the volume of multicellular spheroids.