How does pixmetre calculate the spore ranges? Are they reliable ?

Es gibt 15 Antworten in diesem Thema, welches 3.476 mal aufgerufen wurde. Der letzte Beitrag () ist von Steve_mt.

    I include this example with 5 spore measurements only so that I keep things simple.


    As you see the smallest spore is 6.9 and the largest is 8.2 yet Piximetre ranges the spore sizes 6.2-8.7 and I think the extra 1um (0.5um per end) (c. 18%) is a bit too much erroneous. Should it be something like 6.9 [7.1; 7.7] 8.2 ?!?! I never thought about but when I saw this I was wondering....



    6.93 4.87
    7.08 5.15
    7.89 5.37
    7.12 5.58
    8.18 5.26
    6.2 [6.9; 8] 8.7 × 4.7 [5; 5.5] 5.8 err
    Q = 1.2 [1.3; 1.5] 1.6; N = 5; C = 95%
    Me = 7.4 × 5.2 err; Qe = 1.4

    Hi Steve, I just came across your post from oktober… I also noticed this, when i measured only few spores and remebered their size. I don t think the makers guessed it arbitrary, and suppose they did some comparison with analogue measurements to define the width. Yet i also think for measurements with an 100x objective it could be a bit much. (It uses the same „extra“ range for all objectives- or can one atribute it seperately?)

    I just downloaded the manual and will try to make it through with my rudimentary french, will tell you if i find somerhing out.


    Another thing i noticed was that measurements in the periphery tend to be a bit larger than in the center. I guess it might be due to the sphaerical abberation of the objective lenses, as the pixel density on the cam sensor sholuld be equaly distributed.

    So i think i is more accurate to measure in the center of two or three pictures to get a reliable n= than measure like 20 items on one picture.


    Kind regards

    Ingo


    Update: I found the Formula how spore range is calculated and some other interesting things here in the 3rd pragraph: in short, as I understand it, the correction exceeding min and max you noticed should be a compensation for the lacking gaussian normal distribution when examining only a little part of the total spore print* and is calculated out of the value variability of your actual measurements (I dont know if this is the mathematically correct term). It thus is not a constant (eg. 0.5µm). Unfortunately there seems to be no option to adjust this foramula within piximetre, if you think it exceeds the good.


    *(maybe if you select a part farther in the back of your slide spores are more mature and thus bigger and so on.. and you dont selcect spores to measure arbitrary on your slide but you take a picture of say 40 spores side by side in the same area)


    I hope this sheds some light on whats going on, at least

    Cheers Ingo

    Dear Ingo,


    Thanks for going into this and writing back. I greatly appreciate you did this and have some answers. So as I see, these 'erroneous limits' (Min-Max) are a result of measuring a few spores but should be more reliable when measuring say > 20 as I usually do. Hmm, it is strange that it gives this weird reading to be honest! They should fix it without doing 'assumptions' and error corrections. Knowing this, it is a bit reassuring that the formulas by Pixmetre are fine to report in papers/literature.


    With regards your other point, I was also noticing that spores located at about 10% border of the image are a bit stretched out giving false larger sizes. I hence measure spores around the centre and avoid the margins. I take several images when few spores are present.


    Thanks Ogni

    Hi guys,


    the manual of piximetre explains in very much detail how the statistics is done, see here. So the limits are by no means erroneous, you just have to interpret in the right way. Concerning the use of spore statistics of Piximetre in papers, it is crucial that the author of the paper understands what Piximetre is doing and explains it in the paper in such a way that the reader can also understand the approach used. Otherwise, one will just produce crap instead of good science.


    Björn

    Hi Björn, I am a little disappointed by your comment I must admit. Before writing my comment I thougt about instead posting your name here to call you, as you are to my knowledge the one with most experience in the mathematics/physics field here in this forum. I didnt because I thought if you would have liked to answer you probably would have done so in the meantime und thus either are busy with other things or simply didnt want to, which also is ok. So actually I took the time to look it up and tried to explain it in a simple way, as I understood it (not beeing an mathematician and not talking english after highschool years).

    The term is of course not erroneous (neither is it "right") it is an approximation on reality as one does not want (and cannot) measure all spores of a fruitbody plus the error of the measurement itself. So all we can do (and Piximetre does) is give a certain probability (e.g 95%) that spore sizes are within certain limits.

    For you reading these statistical terms might be like reading soccer charts for someone else, yet for people not involved in this field it might be hard to understand. Which of course is due to insufficient education of statistics at school, but thats another problem. So what I am a bit sad about is you posting the link to the statistics (again) and write that if one does not understand it one should leave it or it will be crap. Which is ok as an opinion but doesnt get us any further and also I think this forum lives from sharing knowledge. I speculate you probably are confronted with crap statistics in papers everyday making you angry which led you saying so. I just had wished you let us participate a little of your knowledge and tell Steve something like "look thats the reason why these limits are correct even if they seem large" or at least "read this chapter on statistics in Wikipedia it might help you understand the term".

    Kind regards, Ingo

    Hi Ingo,


    there are two things that one should distinguish. One is the use of Piximetre by amateurs like you and me who use it to measure spores and then get to some ID that is (hopefully) correct. In this case, one does not have to worry too much about the details of the statistics provided one measures a sufficiently large number of spores (whatever that number is but typically 20 spores should be ok). However, when you want to publish in a scientific journal, you have to be clear about the methods that you use. If you as an author don't know what the programme that you used for measurement does, nobody else will be able to verify of falsify your observations. It is in this case that one should know whether Piximetre calculates deciles and uses them to display some result of the measurements or whether it calculates averages, standard deviations and maximal/minimal spore sizes.


    Björn

    Yes Björn, I am absolutely with you there and this of course is not the intention of this forum, thats what there are lectures for.

    Yet also as an amateur it would be nice to have a basic idea of whats going on;) so I am not so easily pulled over the table... Ok, ok I better shut my mouth now. Good night,

    Ingo

    Thanks Bjorne. The literature in the manual is quite heavy for me to understand and the french - English translation is not so good for this technical doc.


    the format a [b ; c] z states that the min-max range is b - c for covering 80% sample(? did I understand well?) and a - z (mini / maxi) is more or less an estimated 100% range. Hence with a small sample size, the estimation on the mini/maxi may appear out of proportion.


    It would be great if Pixmetere can allow formatting of the formulas. I really like a simpler (min-) mean (-max)


    LG

    Steve

    Zitat von ogni volta

    I just had wished you let us participate a little of your knowledge and tell Steve something like "look thats the reason why these limits are correct even if they seem large" or at least "read this chapter on statistics in Wikipedia it might help you understand the term".

    Kind regards, Ingo

    It is a bit true Ingo. Sometimes, I ask for help and instead of two hands or a hand I get a finger, and that could sometimes be also good enough to help myself and try to understand. Usually, I ask for the second finger (as I did here) and hope I get some reply to my simplified hypothesis. Thanks for your support ;)

    Zitat von Steve_mt

    It would be great if Pixmetere can allow formatting of the formulas. I really like a simpler (min-) mean (-max)

    That is possible, Steve! it is called "formule classique" and you can change it in the menu, where it says "formule dimensionelle" , clic on it and it changes to "formule classique" :)

    (Statistically it is less exact though, as explained in the manual)

    LG Ingo

    Hi guys,


    from my understanding of the manual (as you already said, it's in French after all), the Min [m ; M] Max provides the following information:

    The interval [Min, Max] is centered around the average of the sample and there is a 95% probability (at least if you use the standard settings of Piximetre) that a random spore will have a size inside this interval.

    The interval [m, M] tells us that there is a 95% (standard settings) probability that the average spore size is inside this interval.


    To come back to my original argument: Obviously, one can spend more or less time thinking about the statistics of spore measurements (if I remember correctly, there was an older, very long and detailed thread on this topic in the forum). The important point for scientific publications is that one should be 100% sure about the method that one has used. That means that if one is in doubt one should rather calculate averages, standard deviations, minimal & maximal values etc. by oneself and state the used approach in the method section of the publication.


    Björn

    1. The interval [Min, Max] is centred around the average of the sample and there is a 95% probability that a random spore will have a size inside this interval.So 95% of the sample have spores within this range; some (5%) larger or smaller

    OK

    2. The interval [m, M] tells us that there is a 95% probability that the average spore size is inside this interval.

    This I haven't understood is it like a range of an average? ??

    Hi Steve,


    you have to distinguish two different average values. The first one is the average of the sample that you have measured. That's easy and just given by 1/N*sum_i l_i where l_i is the length of spore i and you sum overall spores from 1 to N. The second average is the average to the true distribution, i.e. if you could measure all spores of all specimens of a given species. Obviously, one cannot determine this value by a direct measurement, but one can infer it using statistical methods which implies that one usually ends up with a certain interval in which this average is with a certain probability. It is this second value which is indicated by the range [m, M].


    Björn