Computer vision-based analytical chemistry applied to determining iron in commercial pharmaceutical formulations

Abstract

Two different computer vision-based analytical chemistry (CVAC) methods were developed to quantify iron in the commercial pharmaceutical formulations Ferbisol and Ferro sanol. The methods involve using a digital camera or a desktop scanner to capture a digital image of a series of Fe standard solutions and the unknown sample upon reaction with o-phenanthroline. The images are processed with appropriate software (e.g., the public domain programme ImageJ, from NIH) to obtain a numerical value (analytical signal) based on colour intensity. The fact that such a value is proportional to the analyte concentration allows one to construct a calibration graph from the standards and interpolate the value for the sample in order to determine its concentration. The results thus obtained were compared with those provided by a spectrophotometric method and the US Pharmacopoeia's recommended method. The differences never exceeded 2%. The two proposed methods are simple and inexpensive; also, they provide an effective instrumental alternative to spectrophotometric methods which can be especially beneficial in those cases where purchasing and maintaining a spectrophotometer is unaffordable.

Chemometrics‐assisted color histogram‐based analytical systems

“…It is important to note that the most CVAC approaches use RGB histogram to calculate several parameters from the first‐order statistics (e.g. mean, mode, and median of the red, green and blue channels), which can be individually used to perform simple linear regression 7,8 or even combined to construct multivariate regression models 13…”

“…It is important to note that the most CVAC approaches use RGB histogram to calculate several parameters from the first-order statistics (e.g. mean, mode, and median of the red, green and blue channels), which can be individually used to perform simple linear regression 7,8 or even combined to construct multivariate regression models. 13 Lyra et al 13 developed and validated calibration models by multiple linear regression (MLR) for the simultaneous determination of sodium and calcium in powder milk.…”

“…The molecular absorption phenomenon (as classically done in the well‐known UV‐Vis spectrophotometry) was indirectly explored because digital images contain the color(s) of the emergent radiation, which is a complementary color of the absorbed radiation by the absorbent molecular species. A recent application was successfully performed for iron (II) determination in pharmaceutical formulations based on its well‐known chelation with o ‐phenanthroline (at pH 3.5) to form a reddish orange complex 7 . For this, they proposed two digital image‐based procedures using a digital camera and a desktop scanner.…”

“…These two abovementioned studies 7,8 are typical examples of computer vision‐based analytical chemistry (CVAC), 3,5 which generally requires several steps for image processing, such as image segmentation, texture analysis, and/or other mathematical correction operations before statistical or chemometric analysis. Image segmentation consists of the partitioning of a digital image into multiple regions (set of pixels) with no overlapping and is typically used to locate objects and boundaries (lines, curves, etc.)…”

“…1,2,6 Therefore, digital imaging has been a good alternative in recent decades in analytical chemistry, because it presents intrinsic advantages of being easy to use, rapid, versatile, objective, low cost, and sometimes noninvasive. [3][4][5][6][7][8] In such scenario, this review systematizes for the first time the fundamentals of exploiting color histograms (obtained from digital images captured with simple commercial devices) as multivariate analytical information in digital image-based chemical analysis. In Section 2, a comprehensive discussion of the different modalities of DIBanalytical approaches derived from color histograms as well as their acronyms will be presented.…”